100% found this document useful (3 votes)

2K views

Distributed Programming With Java

Uploaded by

Raghavendiran J M

Available Formats

Download as PDF, TXT or read online on Scribd

100% found this document useful (3 votes)

2K views

Distributed Programming With Java

Uploaded by

Raghavendiran J M

Available Formats

Download as PDF, TXT or read online on Scribd

You are on page 1/ 418

Distributed Programming With Java

Technology

SL-301

Student Guide With Instructor Notes

Sun Microsystems Inc.

MS BRM01-209
500 Eldorado Blvd.
Broomfield, Colorado 80021
U.S.A.

Revision C.1, March 2000

Copyright © 2000 Sun Microsystems, Inc. 901 San Antonio Road, Palo Alto, California 94303, U.S.A. All rights reserved.
This product or document is protected by copyright and distributed under licenses restricting its use, copying,
distribution, and decompilation. No part of this product or document may be reproduced in any form by any means
without prior written authorization of Sun and its licensors, if any.
Third-party software, including font technology, is copyrighted and licensed from Sun suppliers.
Sun, Sun Microsystems, the Sun Logo, Java, Java 2, JDK, Solaris, JavaSoft, JDBC, JavaBeans, Write Once Run Anywhere,
100% Pure Java, and Java Web Server are trademarks or registered trademarks of Sun Microsystems, Inc. in the U.S. and
other countries.
All SPARC trademarks are used under license and are trademarks or registered trademarks of SPARC International, Inc.
in the U.S. and other countries. Products bearing SPARC trademarks are based upon an architecture developed by Sun
Microsystems, Inc.
Netscape Navigator is a trademark of Netscape Communications Corporation.
UNIX is a registered trademark in the U.S. and other countries, exclusively licensed through X/Open Company, Ltd.
The OPEN LOOK and Sun Graphical User Interface was developed by Sun Microsystems, Inc. for its users and licensees.
Sun acknowledges the pioneering efforts of Xerox in researching and developing the concept of visual or graphical user
interfaces for the computer industry. Sun holds a non-exclusive license from Xerox to the Xerox Graphical User Interface,
which license also covers Sun’s licensees who implement OPEN LOOK GUIs and otherwise comply with Sun’s written
license agreements.
U.S. Government approval required when exporting the product.
RESTRICTED RIGHTS: Use, duplication, or disclosure by the U.S. Govt is subject to restrictions of FAR 52.227-14(g)
(2)(6/87) and FAR 52.227-19(6/87), or DFAR 252.227-7015 (b)(6/95) and DFAR 227.7202-3(a).
DOCUMENTATION IS PROVIDED "AS IS" AND ALL EXPRESS OR IMPLIED CONDITIONS, REPRESENTATIONS,
AND WARRANTIES, INCLUDING ANY IMPLIED WARRANTY OF MERCHANTABILITY, FITNESS FOR A
PARTICULAR PURPOSE OR NON-INFRINGEMENT, ARE DISCLAIMED, EXCEPT TO THE EXTENT THAT SUCH
DISCLAIMERS ARE HELD TO BE LEGALLY INVALID.

Please
Recycle
Contents
About This Course.................................................................................... xiii
Course Goal ....................................................................................... xiii
Course Overview .............................................................................. xiv
Course Map......................................................................................... xv
Module-by-Module Overview ........................................................ xvi
Course Objectives........................................................................... xviii
Skills Gained by Module.................................................................. xix
Guidelines for Module Pacing ......................................................... xx
Topics Not Covered.......................................................................... xxi
How Prepared Are You?................................................................. xxii
Introductions .................................................................................. xxiii
How to Use Course Materials ...................................................... xxiv
Course Icons and Typographical Conventions ......................... xxvi
Icons ......................................................................................... xxvi
Typographical Conventions ................................................ xxvii
Notes to the Instructor.................................................................... xxix
Philosophy ...............................................................................xxix
Course Tools ............................................................................xxix
Instructor Setup Notes .................................................................. xxxii
Overview of Distributed Computing.....................................................1-1
Objectives ............................................................................................1-1
Relevance............................................................................................ 1-2
Additional Resources ....................................................................... 1-3
History of Computing ...................................................................... 1-4
Distributed Computing.................................................................... 1-6
Importance of Distributed Computing ..................................1-7
Characteristics of Distributed Computing ............................1-8
Design Considerations ..................................................................... 1-9
Latency........................................................................................1-9
Partial Failure ............................................................................1-9
Distributed Programming Architectures..................................... 1-11
Overview of Daemon Processes ...........................................1-12
Using Java Technologies to Implement Daemons .............1-13

iii
Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
Third-Party Java Technologies Using Daemons ................1-15
Understanding Daemons.......................................................1-15
Overview of Remote Procedure Call ...................................1-16
Third-Party Java Technologies Using RPC .........................1-17
Understanding Remote Procedure Call (RPC) ...................1-17
Overview of Remote Objects.................................................1-18
Implementing Remote Objects With Java Technologies ...1-19
Understanding Remote Objects ............................................1-20
Overview of Object Bus Systems ..........................................1-21
JavaBeans InfoBus...................................................................1-22
Understanding Object Bus Systems .....................................1-23
Overview of Mobile Agents ..................................................1-24
Understanding Mobile Agents..............................................1-26
Supporting Technologies for Distributed Computing .............. 1-27
Naming Service .......................................................................1-28
Security Service .......................................................................1-28
Transaction Service .................................................................1-28
Event Service............................................................................1-29
Message Queueing..................................................................1-29
Supporting Technologies .......................................................1-30
Check Your Progress ...................................................................... 1-31
Think Beyond .................................................................................. 1-32
Java Database Connectivity (JDBC) .......................................................2-1
Objectives ............................................................................................2-1
Relevance............................................................................................ 2-2
Additional Resources ....................................................................... 2-2
Introduction ....................................................................................... 2-3
JDBC Drivers and Driver Managers............................................... 2-4
java.sql Package ............................................................................ 2-7
JDBC Flow........................................................................................ 2-10
JDBC Programmer’s Interface ....................................................... 2-11
JDBC Example .........................................................................2-12
Exercise: Compiling a JDBC Application .................................... 2-14
Preparation...............................................................................2-14
Tasks .........................................................................................2-14
Exercise Summary...................................................................2-15
Using JDBC Drivers ................................................................2-16
Exercise: Loading a Driver............................................................. 2-18
Tasks .........................................................................................2-18
Exercise Summary...................................................................2-19
Exercise: Loading Driver 2 (Optional) ......................................... 2-22
Tasks .........................................................................................2-22
Specifying a Database.............................................................2-24
NIS Name Resolution Example ............................................2-26
Opening a Database Connection ..........................................2-27

iv Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
Exercise: Connecting to a Database.............................................. 2-29
Tasks .........................................................................................2-29
Exercise Summary...................................................................2-30
Submitting a Query ................................................................2-31
Using a Prepared Statement ..................................................2-33
Using Callable Statements .....................................................2-36
Receiving Results ....................................................................2-38
Exercise: Executing an SQL Query ............................................... 2-40
Tasks .........................................................................................2-40
Exercise Summary...................................................................2-41
JDBC Driver Architecture .............................................................. 2-44
JDBC Driver Categories .........................................................2-44
Application Designs ....................................................................... 2-46
Two-Tier Application Designs ..............................................2-46
Three-Tier Application Designs............................................2-47
Applets.............................................................................................. 2-48
Applets and Traditional Database Applications ................2-49
Performance Considerations .................................................2-49
Security Limitations................................................................2-49
Exercise: Building a JDBC Application (Optional)..................... 2-50
Preparation...............................................................................2-50
Tasks .........................................................................................2-50
Exercise Summary...................................................................2-52
More Information............................................................................ 2-53
Check Your Progress ...................................................................... 2-54
Think Beyond .................................................................................. 2-55
Remote Method Invocation (RMI) .........................................................3-1
Objectives ............................................................................................3-1
Relevance............................................................................................ 3-2
Additional Resources ....................................................................... 3-4
What Is Java RMI?.....................................................................3-5
RMI Characteristics...................................................................3-7
RMI Application Architecture.................................................3-9
What Is Serialization? .............................................................3-10
Object Serialization Architecture ..........................................3-12
Writing and Reading to and from an Object Stream .........3-13
Object Streams .........................................................................3-16
The Externalizable Interface.............................................3-18
Exercise: Serialization..................................................................... 3-19
Tasks .........................................................................................3-19
Exercise Summary...................................................................3-21
Creating an RMI Application ........................................................ 3-22
Steps to Create an RMI Application.....................................3-23
Deploying an RMI Application.............................................3-29

v
Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
Exercise: Compiling a Basic RMI Application ............................ 3-31
Tasks .........................................................................................3-31
Exercise Summary...................................................................3-33
RMI Architecture............................................................................. 3-34
RMI Architecture Overview ..................................................3-34
Invocation Overview ..............................................................3-36
Interface Descriptions.............................................................3-38
The Application Layer............................................................3-39
The Stub and Skeleton Layer.................................................3-40
Stub Communication..............................................................3-41
Skeleton Communication.......................................................3-42
The Remote Reference Layer.................................................3-43
The Transport Layer ...............................................................3-44
Garbage Collection .................................................................3-46
RMI Object Hierarchy.............................................................3-49
The rmiregistry Application..............................................3-53
Remotely Loaded Code.................................................................. 3-56
The java.rmi.server.codebase Property ......................3-57
Security Aspects ......................................................................3-58
Exercise: Remotely Loaded Code ................................................. 3-59
Preparation...............................................................................3-59
Tasks .........................................................................................3-59
Exercise Summary...................................................................3-61
Object Factories .......................................................................3-62
Exercise: Object Factory ................................................................. 3-66
Tasks .........................................................................................3-66
Exercise Summary...................................................................3-68
Object Activation.....................................................................3-69
Exercise: Object Activation ............................................................ 3-74
Tasks .........................................................................................3-74
Exercise Summary...................................................................3-75
Objects as Parameters in Remote Calls ................................3-76
RMI “Agents” ..........................................................................3-76
Exercise: Objects as Parameters in Remote Calls ....................... 3-80
Preparation...............................................................................3-80
Tasks .........................................................................................3-80
Exercise Summary...................................................................3-82
HTTP Tunneling......................................................................3-83
Exercise: Developing an RMI Application From
Scratch (Optional) ........................................................................ 3-85
Tasks .........................................................................................3-85
Exercise Summary...................................................................3-87
Check Your Progress ...................................................................... 3-88
Think Beyond .................................................................................. 3-89

vi Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
Java Interface Definition Language (JavaIDL).....................................4-1
Objectives ............................................................................................4-1
Relevance............................................................................................ 4-2
Additional Resources ....................................................................... 4-3
Module Overview ............................................................................. 4-4
Object Request Broker ...................................................................... 4-5
ORB Implementation........................................................................ 4-7
Static and Dynamic Invocation ...............................................4-7
Interface Repository..................................................................4-8
Object Adapter ..........................................................................4-8
CORBAservices .........................................................................4-9
CORBAfacilities.........................................................................4-9
Wrapping Legacy Code With CORBA ........................................ 4-12
JavaIDL in Relation to CORBA ..................................................... 4-14
JavaIDL – A Full ORB? ...........................................................4-14
Is a Commercial ORB Necessary?.........................................4-15
Interface Definition Language Basics........................................... 4-17
IDL Language Mappings .......................................................4-19
Important IDL Keywords ......................................................4-20
JavaIDL Architecture Overview ................................................... 4-22
Creating and Deploying a JavaIDL Application ........................ 4-24
Creating a JavaIDL Application............................................4-24
Deploying a JavaIDL Application ........................................4-30
Exercise: Compiling an Application............................................. 4-32
Tasks .........................................................................................4-32
Exercise Summary...................................................................4-33
Bootstrapping the JavaIDL System .............................................. 4-34
What Is Bootstrapping?..........................................................4-34
Bootstrapping the Client Application ..................................4-35
Exercise: Bootstrapping/COS Name Server ............................... 4-39
Preparation...............................................................................4-39
Tasks .........................................................................................4-39
Exercise Summary...................................................................4-41
IDL-to-Java Programming Language Mapping Details ............ 4-42
Module Construct ...................................................................4-43
Interface Construct..................................................................4-44
Interface Example ...................................................................4-44
Operations and Parameter Declarations..............................4-46
Attribute Declaration..............................................................4-48
Raises Expressions and Exceptions ......................................4-50
The typedef Keyword...........................................................4-52
Basic Java Technology Types ................................................4-53
The struct Keyword...........................................................4-54
The Sequence Keyword.........................................................4-56
Sequence Mapping..................................................................4-57

vii
Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
Array .........................................................................................4-58
The enum Construct ...............................................................4-59
Exercise: IDL-to-Java Programming Language Mapping
Details ............................................................................................ 4-61
Tasks .........................................................................................4-61
Exercise Summary...................................................................4-64
Objects as Parameters in Remote Calls ........................................ 4-65
Exercise: Objects as Parameters in Remote Calls ....................... 4-67
Tasks .........................................................................................4-67
Exercise Summary...................................................................4-69
Futures ......................................................................................4-71
Check Your Progress ...................................................................... 4-73
Think Beyond .................................................................................. 4-74
Servlets.........................................................................................................5-1
Objectives ............................................................................................5-1
Relevance............................................................................................ 5-2
Additional Resources ....................................................................... 5-3
Servlets Overview ............................................................................. 5-4
Servlets API........................................................................................ 5-5
The javax.servlet Package..................................................5-6
Simple Servlet ............................................................................5-7
Servlet Interaction .....................................................................5-8
HTTP Servlets .................................................................................... 5-9
The javax.servlet.http Package.....................................5-10
HTTP Servlet Example ...........................................................5-12
Exercise: Creating Simple HTTP Servlets.................................... 5-14
Preparation...............................................................................5-14
Tasks .........................................................................................5-14
Using an HTTP Servlet...........................................................5-17
Example Usage ........................................................................5-18
Servlet Life Cycle ............................................................................ 5-19
HTTP Servlet Request .................................................................... 5-20
Exercise: Snoop Servlet................................................................... 5-22
Tasks .........................................................................................5-22
Exercise Summary...................................................................5-23
HTTP Servlet Response..........................................................5-24
HTTP Session Management ..................................................5-25
Exercise: Session Servlet................................................................. 5-26
Preparation...............................................................................5-26
Tasks .........................................................................................5-26
Exercise Summary...................................................................5-27
Application Designs ....................................................................... 5-28
Two-Tier Design......................................................................5-29
Three-Tier Design ...................................................................5-30

viii Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
Exercise: LeShop Servlet (Optional) ............................................. 5-31
Preparation...............................................................................5-31
Tasks .........................................................................................5-31
Exercise Summary...................................................................5-32
Check Your Progress ...................................................................... 5-33
Think Beyond .................................................................................. 5-34
Object Bus Systems ...................................................................................6-1
Objectives ............................................................................................6-1
Relevance............................................................................................ 6-2
The Object Bus Model ...................................................................... 6-3
Object Bus Example .......................................................................... 6-4
Object Bus Compared to CORBA and RMI .................................. 6-5
Object Bus Architectures.................................................................. 6-7
Hub-and-Spoke Architecture ..................................................6-7
Multicast Bus Architecture ......................................................6-9
Application Areas Suitable for Object Bus .................................. 6-11
Products and Standards ................................................................. 6-12
SoftWired iBus................................................................................. 6-13
Sample iBus Application................................................................ 6-14
The Talker Program ................................................................6-14
The Listener Program .............................................................6-16
Exercise: iBus ................................................................................... 6-18
Tasks .........................................................................................6-18
Exercise Summary...................................................................6-19
The iBus API .................................................................................... 6-20
The Stack ..................................................................................6-20
Channels and URLs ................................................................6-22
Posting Objects ........................................................................6-24
Exercise: Creating a Stock Quote Application ............................ 6-26
Tasks .........................................................................................6-26
Exercise Summary...................................................................6-27
Check Your Progress ...................................................................... 6-28
Think Beyond .................................................................................. 6-29
Supporting Technologies .........................................................................7-1
Objectives ............................................................................................7-1
Relevance............................................................................................ 7-2
Overview ............................................................................................ 7-3
Java Naming and Directory Interface ............................................ 7-4
Naming Services and Directory Services....................................... 7-6
Naming Services........................................................................7-6
Directory Services .....................................................................7-7
Composite Names............................................................................. 7-8
Architecture Overview..................................................................... 7-9
Service Providers............................................................................. 7-10

ix
Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
Module Exercise Overview ........................................................... 7-11
Using JNDI to Access COS Naming.....................................7-12
Exercise: Java Naming and Directory Interface.......................... 7-13
Preparation...............................................................................7-13
Tasks .........................................................................................7-13
Java Transaction Service ................................................................ 7-16
Java Message Service ...................................................................... 7-17
JMS Service Provider ..............................................................7-18
JMS Domains ...........................................................................7-18
JMS Domains – Point-to-Point ...................................................... 7-19
JMS Domains – Publish and Subscribe ........................................ 7-20
Check Your Progress ...................................................................... 7-21
Think Beyond .................................................................................. 7-22
Technology Summary and Comparison................................................8-1
Objectives ............................................................................................8-1
Relevance............................................................................................ 8-2
Complementary or Overlapping Technology?............................. 8-3
JDBC Usage........................................................................................ 8-5
Using Servlets .................................................................................... 8-6
Example of an n-Tier Architecture ................................................. 8-7
RMI Compared to JavaIDL.............................................................. 8-8
JavaIDL Advantages......................................................................... 8-9
JavaIDL Disadvantages.................................................................. 8-11
Java RMI Advantages..................................................................... 8-13
Java RMI Disadvantages ................................................................ 8-14
RMI Over IIOP................................................................................. 8-15
RMI and JavaIDL ............................................................................ 8-16
Request-Reply Compared to Publish-Subscribe ........................ 8-17
Bus Example 1 – Real-Time Enterprise ........................................ 8-19
Bus Example 2 – Flat Architecture................................................ 8-20
Check Your Progress ...................................................................... 8-21
Think Beyond .................................................................................. 8-22
Flags for the idltojava Utility ..............................................................A-1
The IDL-to-Java Compiler: idltojava......................................... A-2
Syntax ........................................................................................A-2
Description................................................................................A-2
Options ......................................................................................A-2
Flags ...........................................................................................A-3
Using the #pragma compiler directive..................................A-5

x Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
SQL Syntax................................................................................................. B-1
SQL Commands ............................................................................... B-2
SELECT Statement............................................................................. B-3
Syntax ........................................................................................ B-3
Examples ................................................................................... B-4
Joins............................................................................................ B-5
INSERT Statement............................................................................. B-6
DELETE Statement............................................................................. B-7
UPDATE Statement............................................................................. B-8
Glossary ......................................................................................... Glossary-1

xi
Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
About This Course

Course Goal
Distributed Programming With Java Technology provides you with the
knowledge and skills necessary to program distributed computing
applications using the distributed technologies from JavaSoft™:

● Java™ Database Connectivity (JDBC™) application programming

interface (API)

● Remote Method Invocation (RMI) API

● JavaIDL (Java interface definition language) API

● Java technology servlets (Java servlets)

You will also learn about the supporting technologies, such as Java
Transaction Services (JTS), Java Naming and Directory Interface
(JNDI), and Java Message Service (JMS). At the end of the course, the
knowledge gained should enable you to make informed decisions
about which technology is best used under which circumstances.
✓ Use this module to get the students excited about this course.
✓ With regard to the overheads: To avoid confusion among the students, it is very important
to tell them that the page numbers on the overheads have no relation to the page numbers
in their course materials. They should use the title of each overhead as a reference.
✓ The strategy provided by the “About This Course” is to introduce students to the course
before they introduce themselves to you and one another. By familiarizing them with the
content of the course first, their introductions will have more meaning in relation to the
course prerequisites and objectives.
✓ Use this introduction to the course to determine how well students are equipped with the
prerequisite knowledge and skills. The pacing chart on xx enables you to determine what
adjustments you need to make in order to accommodate the learning needs of students.

xiii
Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
Course Overview

This course is concerned with providing you with the skills and
concepts necessary to solve distributed computing problems using the
four distributed computing technologies from JavaSoft:

● JDBC

● RMI

● JavaIDL

● Java servlets

Issues related to distributed programming, such as a naming or

transaction service, persistent objects, or security issues with remotely
loaded code, are also addressed in this course.

xiv Distributed Programming With Java Technology

The following course map enables you to see what you have
accomplished and where you are going in reference to the course goal.

Overview

Overview of
Distributed Computing

Java Technology
Java Database Remote Method
Connectivity (JDBC) Invocation (RMI)

Java Interface Definition Servlets

Language (JavaIDL)

Push Technology

Object Bus
Systems

Supporting Technology

Supporting
Technologies

Summary

Technology Summary
and Comparison

About This Course xv

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
Module-by-Module Overview

This course contains the following modules:

● Module 1 – “Overview of Distributed Computing”

This module describes the characteristics of distributed computing

environments and of distributed object computing. Various
techniques used to solve distributed computing problems are
presented, with an overview of the JavaSoft technologies: JDBC,
RMI, JavaIDL and Java servlets.

● Module 2 – “Java Database Connectivity (JDBC)”

This module describes the main features of JDBC, the Java

technology classes (Java classes) provided by JDBC, and how to
use the JDBC to interface with a database system.

xvi Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
● Module 3 – “Remote Method Invocation (RMI)”

This module describes the main features of RMI, the classes

provided by RMI, and how to use RMI to solve distributed
computing problems. The new features of RMI that are delivered
with the Java 2 SDK, standard edition, Version 1.2 technology
are also presented.

● Module 4 – “Java Interface Definition Language (JavaIDL)”

This module describes the Common Object Request Broker

Architecture (CORBA), the main features of JavaIDL, how to use
JavaIDL to solve distributed computing problems, and
information about the IDL-to-Java programming language
mapping.

● Module 5 – “Servlets”

This module describes generic Java servlets and hypertext transfer

protocol (HTTP) servlets.

● Module 6 – “Object Bus Systems”

This module describes object bus systems. Using object bus

systems, you can build systems using a multicast, many-to-many
object communication paradigm.

● Module 7 – “Supporting Technologies”

This module describes three supporting technologies for doing

distributed computing.

● Module 8 – “Technology Summary and Comparison”

This module summarizes the different technologies explained in

this course, and compares them to each other.

About This Course xvii

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
Course Objectives

Upon completion of this course, you should be able to:

● Describe the basics of distributed computing technologies

● Write a JDBC applet or application

● Write an RMI applet or application

● Write a JavaIDL applet or application

● Write a Java technology-based servlet

● Explain how object bus systems, publish-subscribe systems, and

remote events work

● Explain the basics of JTS, JNDI, and JMS

● Compare and contrast the three distributed computing Java

technologies available from Sun Microsystems™

xviii Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
Skills Gained by Module

The skills for Distributed Programming With Java Technology are shown
in the first column of the matrix below. The black boxes indicate the
main coverage for a topic; the gray boxes indicate that the topic is
briefly discussed.

Module

Skills Learned 1 2 3 4 5 6 7 8

Describe the basics of distributed computing technologies

Write a JDBC applet or application
Write an RMI applet or application
Write a JavaIDL applet or application
Write a Java servlet
Explain how object bus systems, publish-subscribe, and remote
events work
Explain the basics of JTS, JNDI, and JMS
Compare and contrast the three distributed computing Java
technologies available from Sun Microsystems

✓ Refer students back to this matrix as you progress through the course to show them the
progress they are making in learning the skills advertised for this course.

About This Course xix

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
Guidelines for Module Pacing

The following table provides a rough estimate of pacing for this

course.

Module Day 1 Day 2 Day 3 Day 4 Day 5

“About This Course” A.M.

Module 1 – “Overview of Distributed A.M.
Computing”
Module 2 – “Java Database P.M. A.M.
Connectivity (JDBC)”
Module 3 – “Remote Method P.M. A.M.
Invocation (RMI)”
Module 4 – “Java Interface Definition P.M. A.M.
Language (JavaIDL)”
Module 5 – “Servlets” P.M.
Module 6 – “Object Bus Systems” A.M.
Module 7 – “Supporting A.M.
Technologies”
Module 8 – “Technology Summary P.M.
and Comparison”

✓ This table should only be used as a guideline for timing throughout the week. The time
you spend on each module and lab may differ from class to class, as knowledge and Java
programming language experience of the students varies.

xx Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
Topics Not Covered

This course does not cover the topics shown on the above overhead.
Many of the topics listed on the overhead are covered in other courses
offered by Sun Educational Services. Refer to the Sun Educational
Services catalog for specific information and registration.

About This Course xxi

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
How Prepared Are You?

To be sure you are prepared to take this course, you should be able to
do the following:

● Write general Java technology-based applications

● Construct simple structured query language (SQL) queries to

obtain database information

● Construct simple data manipulation statements to insert, update,

or delete data in a database

● Describe the basic CORBA architecture

✓ If any students indicate they cannot do the above, meet with them at the first break to
decide how to proceed with the class. Do they want to take the class at a later date? Is
there some way to get extra help?
✓ It might be appropriate here to recommend resources from the Sun Educational Services
catalog that provide training for topics not covered in this course.

xxii Distributed Programming With Java Technology

Now that you have been introduced to the course, introduce yourself
to each other and the instructor, addressing the items shown on the
above overhead.

About This Course xxiii

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
How to Use Course Materials

To enable you to succeed in this course, these course materials employ

a learning model that is composed of the following components:

● Course map – An overview of the course content appears in the

"About This Course" module so you can see how each module fits
into the overall course goal.

● Objectives - What you should be able to accomplish after

completing this module is listed here.

● Relevance – The relevance section for each module provides

scenarios or questions that introduce you to the information
contained in the module and provoke you to think about how the
module content relates to your interest in learning Java technology
programming (Java programming).

● Overhead image – Reduced overhead images for the course are

included in the course materials to help you easily follow where
the instructor is at any point in time. Overheads do not appear on
every page.

xxiv Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
● Lecture – The instructor presents information specific to the topic
of the module. This information helps you learn the knowledge
and skills necessary to succeed with the exercises.

● Exercise – Lab exercises give you the opportunity to practice your

skills and apply the concepts presented in the lecture.

● Check your progress – Module objectives are restated, sometimes

in question format, so that before moving on to the next module
you are sure that you can accomplish the objectives of the current
module.

● Think beyond – Thought-provoking questions are posed to help

you apply the content of the module or predict the content in the
next module.

About This Course xxv

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
Course Icons and Typographical Conventions

The following icons and typographical conventions are used in this

course to represent various training elements and alternative learning
resources.

Icons

Discussion – Indicates a small-group or class discussion on the current

topic is recommended at this time.

Exercise objective – Indicates the objective for the lab exercises that
follow. The exercises are appropriate for the material being discussed.

Note – Additional important, reinforcing, interesting or special

information.

Caution – A potential hazard to data or machinery.

!
Warning – Anything that poses personal danger or irreversible
damage to data or the operating system.

xxvi Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
Typographical Conventions
Courier is used for the names of commands, files, and directories, as
well as on-screen computer output. For example:

Use ls -al to list all files.

system% You have mail.

It is also used to represent parts of the Java™ programming language

such as class names, methods, and keywords. For example:

The getServletInfo method is used to...

The java.awt.Dialog class contains Dialog (Frame parent)

Courier bold is used for characters and numbers that you type. For
example:

system% su
Password:

It is also used for each code line that will be referenced in text.
For example:

1. import java.io.*;
2. import javax.servlet.*;
3. import javax.servlet.http.*;

Courier italic is used for variables and command-line

placeholders that are replaced with a real name or value. For example:

To delete a file, type rm filename.

Palatino italics is used for book titles, new words or terms, or words
that are emphasized. For example:

Read Chapter 6 in User’s Guide.

These are called class options.
You must be root to do this.

About This Course xxvii

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
The Java programming language examples use the following
additional conventions:

● Method names are not followed with parentheses unless a formal

or actual parameter list is shown. For example:

"The doIT method..." refers to any method called doIt.

"The doIt() method..." refers to a method called doIt which takes

no arguments.

● Line breaks occur only where there are separations (commas),

conjunctions (operators), or white space in the code. Broken code
is indented four spaces under the starting code.

● If a command is different on the Solaris™ Operating Environment

and Microsoft Windows platforms, both commands are shown.
For example:

On Solaris Operating Environment

cd server_root/bin

On Microsoft Windows

cd server_root\bin

xxviii Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
Notes to the Instructor

Philosophy
The Distributed Programming With Java Technology course has been
created to allow for interactions between the instructor and the
student as well as between the students themselves. In an effort to
enable you to accomplish the course objectives easily, and in the time
frame given, a series of tools have been developed and support
materials created for your discretionary use.

A consistent structure has been used throughout this course. This

structure is outlined in the “Course Goal” section. The suggested flow
for each module is:

1. Location of module in the course map

2. Context questions/module rationale
3. Module objectives
4. Lecture information with appropriate overheads
5. Lab exercises
6. Discussion: either as whole class or in small groups

To allow the instructor flexibility and give time for meaningful

discussions during the “relevance” periods, the lectures, and the
small-group discussions, a timing table is included in “General Timing
Recommendations.”

Course Tools
To enable you to follow this structure, the following supplementary
materials are provided with this course:

● Relevance

These questions or scenarios set the context of the module. It is

suggested that the instructor ask these questions and discuss the
answers. The answers are provided only in the instructor guide.

About This Course xxix

The course map allows the students to get a visual picture of the
course. It also helps students know where they have been, where
they are, and where they are going. The course map is presented
in the “About This Course” section in the student guide.

● Lecture overheads

Overheads for the course are provided in two formats:

The paper-based format can be copied onto standard

transparencies and used on a standard overhead projector. These
overheads are also provided in the student’s guide.

The Web browser–based format is in HTML and can be projected

using a projection system which displays from a workstation. This
format gives the instructor the ability to allow the students to view
the overhead information on individual workstations. It also
allows better random access to the overheads.

● Small-group discussion

After the lab exercises, it is a good idea to debrief the students.

You can gather them back into the classroom and have them
discuss their discoveries, problems, and issues in programming
the solution to the problem in small groups of four or five, one-on-
one, or one-on-many.

xxx Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
● General timing recommendations

Each module contains a “Relevance” section. This section may

present a scenario relating to the content presented in the module,
or it may present questions that stimulate students to think about
the content that will be presented. Engage the students in relating
experiences or posing possible answers to the questions. Spend no
more than 10–15 minutes on this section.

Lecture Lab Total Time

Module
(Minutes) (Minutes) (Minutes)
Preface 60 60
Module 1 120 120
Module 2 180 180 360
Module 3 180 180 360
Module 4 180 180 360
Module 5 120 120 240
Module 6 60 60 120
Module 7 60 60 120
Module 8 120 120
Debriefing 120 120

● Module self-check

Each module contains a checklist for students under “Check Your

Progress.” Give them a little time to read through this checklist
before going on to the next lecture. Ask them to see you for items
they do not feel comfortable checking off.

About This Course xxxi

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
Instructor Setup Notes

Purpose of This Guide

This guide provides general information about setting up the
classroom. Refer to the README_setup_Instructions file in the
SL301_IN directory for specific information about how to set up this
course.

Projection System and Workstation

If you have a projection system for projecting HTML slides and are
planning to use the HTML slides, you need to do the following:

● Install the HTML overheads on the workstation connected to the

projection system so you can display them with a browser during
lecture.

To install the HTML overheads on the machine connected to your

overhead projection system, copy the HTML and images
subdirectories provided in the SL301_OH directory to any directory
on the overhead workstation machine.

Display the overheads in the browser by choosing Open ➤ File

and typing the following in the Selection field of the pop-up
window:

/SL301_OH/HTML/OH.Title.doc.html

● Set up an overhead-projection system that can project instructor

workstation screens.

Note – This document does not describe the steps necessary to set up
an overhead projection system because it is unknown what will be
available in each training center. This setup is the responsibility of
each training center.

xxxii Distributed Programming With Java Technology

All of the course files for this course are available from the
education.central server. You can use ftp or the
education.central Web site,
http://education.central/courses/ to download the files from
education.central. Either of these methods requires you to know
the user ID and password for FTP access. See your manager for these
if you have not done this before.

Course Components
This course consists of the following components:

● Instructor guide

The SL301_IG directory contains the FrameMaker files for the

instructor’s guide (student’s guide with instructor notes). The ART
directory is required for printing this guide.

● Student guide

The SL301_SG directory contains the FrameMaker files for the

student’s guide. The ART directory is required for printing this
guide.

● Art

The SL301_ART directory contains the supporting images and

artwork for the student’s and instructor’s guides. This directory is
required for the printing of the student’s and instructor’s guides
and should be located in the same directory as SL301_IG and
SL301_SG.

● Instructor notes

The SL301_IN directory contains the text file

README_setup_Instructions.

About This Course xxxiii

The SL301_OH directory contains the instructor overheads. There

are both HTML and FrameMaker versions of the overheads.

● Lab Files

The SL301_SOL_LF and SL301_WIN_LF directories contain the lab

files for this course.

xxxiv Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
Overview of Distributed
Computing 1

Objectives
Upon completion of this module, you should be able to:

● List the five supporting technologies for distributed computing

● Compare and contrast the different architectures for distributed

computing

● List the distributed computing technologies available for the Java

programming language

Growing popularity and use of intranets and the Internet have

increased the demand for distributed computing applications. This
module provides an overview of distributed computing technologies,
including the four distributed programming technologies available
from the JavaSoft product.

1-1
Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
1
Relevance

✓ Present the following questions to stimulate the students and get them thinking about the
issues and topics presented in this module. They are not expected to know the answers to
these questions. The answers to these questions should be of interest to the students,
and inspire them to learn the content presented in this module.

Discussion – Suppose you have an intranet with a wide range of

hardware and operating environments. You plan to implement a
distributed computing strategy for your company. Consider the
following questions:

● What is your understanding of distributed computing?

✓ This is a very interesting question – pay close attention to the responses, it will help to
define the course for the rest of the week. Some people believe that “distributed” is
synonymous with “parallel” and may define distributed as a single task split over multiple
CPUs. The definition used in this course is different: distributed means having program
resources on multiple address spaces.

● What advantages does distribution offer?

✓ Obvious advantages are better control over the use/development of system resources,
and the ability to share the load of a system over multiple address spaces, but be
careful—keep in mind the previous discussion.

● What existing technologies use distributed computing?

✓ NFS, client-server database systems, SunNet Manager, and so forth.

● What distributed computing technologies do you know?

✓ Daemons, RPC, Distributed Object, Object Bus, JDBC, Servlets, RMI, and so forth.

1-2 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
1
Additional Resources

Additional resources – The following resources can provide

additional detail on the topics presented in this module:

● Orfali, Robert, Bruce Harkey, and Bob Edwards. 1996. The Essential
Distributed Objects Survival Guide. Wiley Press.

● Farley, Jim. 1997. Java Distributed Computing. O’Reilly & Associates.

● A Note on Distributed Computing. [Online]. Sun Microsystems, Inc.

Available:
http://www.sunlabs.com/technical-reports/1994/
abstract-29.html

Overview of Distributed Computing 1-3

History of Computing

The history of computing began with the development of programs

that ran within a single computer; users time-shared the central
processing unit (CPU) resource of a mainframe system and accessed
their “address space” from a terminal.

As computers became more powerful, smaller, and capable of

executing a greater number of programs, the computer moved from
the central location to the desktop. People soon realized there was an
advantage to linking the distributed computers to a network and
sharing common resources (applications) using servers on the
network. With this approach, the client downloads or accesses an
application from a common server and runs the application in the
client’s address space. Overall, this greatly improves the performance
of the application and the client machine is the determining factor in
the speed of the application.

1-4 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
1
However, the client-server environment uncovered issues with the
sharing of common resources—compatibility and size. In a true
homogeneous computing environment, where every client machine is
running the latest operating system, and every machine is of the same
architecture, a single program can be shared from the server. In real
life, there are numerous versions of the operating system (OS), the
hardware architectures are different (even from the same vendor), and
the likelihood is small that a single version of a program can run
across all client machines.

Additionally, applications have grown in size and processing data now

moves down to the client, causing the client programs to be large and
therefore slow to load and start up.

The Java programming language makes it possible to create

applications that run on any platform that supports the Java virtual
machine (JVM), which greatly increases the likelihood that the
application downloaded from the server will run. However, it does not
solve the problem of size; the client program is still large if processing
occurs within the client’s address space.

Overview of Distributed Computing 1-5

Distributed Computing

To solve the problem of program size, you need a way of moving the
major processing pieces to another address space. This would allow
the client to concentrate on the specific pieces required for the end-
user of the application.

The solution is, ironically, somewhere between the single mainframe

and the client-server model. The answer is to allow applications that
are run on the client to access applications that are running on another
machine—a server or another client. This technique is called distributed
computing.

1-6 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
1
Importance of Distributed Computing
Distributed computing is important for several reasons:

● Software development costs are lowered – Many clients can access

the same code; the client-side application development process is
reduced because duplication of effort is reduced.

● Resource loading is better balanced – The whole client-server

paradigm is based on the ability of many machines or processes to
request the resources or services provided by a single or a few
machines or processes, so that CPU, memory, and disk resources
are more effectively allocated.

● Platform independence is possible – The client and server

communicate through a known protocol that does not rely on
hardware.

Overview of Distributed Computing 1-7

Characteristics of Distributed Computing

In a distributed environment, programs that the client runs make calls
to programs in other address spaces, either locally or remotely. This
approach allows the client program to be written specifically for the
client architecture and still allows the client program to make calls
(through a known protocol) to server programs that potentially are
running in a different architecture.

1-8 Distributed Programming With Java Technology

Design Considerations

Latency
Proper analysis of what should reside locally and what can be
accessed remotely is crucial to the performance of the distributed
application. Obviously, a service that can be accessed over a local bus
will be faster than one that must use the network. As the number and
speed of processors that can be put in a system increases, the speed
gap between local and remote access increases.
✓ “Note on Distributed Computing” states that latency can be as high as four to five orders
of magnitude between local and remote access.

Partial Failure
In a local application, if a single component fails (a disk crashes, a
CPU panics, or memory fills up), the application will fail to run. When
writing a local application you do not have to take catastrophic failure
into consideration, because there is little if anything that you can do to
recover.

Overview of Distributed Computing 1-9

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
1
✓ Partial failure dictates that the programs deal with indeterminacy.

However, in the distributed programming model, one component can

fail (the client machine, the server machine, or the network), so partial
failure detection and recovery procedures become important. For
example, if a client application can no longer reach its server process,
should it cease its execution? Should it retry the request? Should it
wait a period of time before a retry? Should it attempt to request the
service from somewhere else, and how many times should it retry
before giving up? From the client side, failure of a processor on the
server might not be distinguishable from a network failure.
✓ The following pages on the NFS™ environment are designed to get the students thinking
about failure and recovery scenarios for distributed applications. Point out that the
systems you will be learning about this week have many parts or components that can
fail; that is, the database host, the database itself, the RMI registry, the Java technology
runtime on the remote servers, the ORB, and so on.

1-10 Distributed Programming With Java Technology

Distributed Programming Architectures

Over time, several approaches to distributed programming have been

developed. While some of the older technologies might become
obsolete (especially Remote Procedure Call), others still have a valid
place today. The choice of a particular technology depends on what
you want to achieve. At the end of this course you should not only
know the different technologies available, but also be able to choose
the most appropriate technology for the specific application that you
want to build. The following approaches are examined in this module:

● Daemons

● Remote Procedure Call (RPC)

● Remote objects

● Object bus systems

● Mobile agents

Overview of Distributed Computing 1-11

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
1
Overview of Daemon Processes
A daemon is a process running on a machine that waits for service
requests, and then services these requests. A local daemon could be a
Java programming language thread (Java thread), which is normally
suspended. The service request activates the thread. In a distributed
environment, a daemon usually listens on an IP port for service
requests, and processes these requests when they come in. Well-known
examples of these types of daemons are:

● ftpd

● telnetd

● httpd

● sendmail

The communication between the client application and the daemon is

usually performed using a daemon-specific protocol, such as Simple
Message Transfer Protocol (SMTP) in the case of the sendmail
daemon.

1-12 Distributed Programming With Java Technology

Using Java Technologies to Implement Daemons

You can develop daemon-like systems using nothing more than Java
technology sockets (Java sockets). In fact, this is what you do when
you develop a Java technology-based application that must interface
with a legacy system. These legacy systems are usually daemons
conceptually with their own proprietary communication protocol.

However, this course covers higher-level APIs, encapsulating the

details of common, standardized types of daemons: database and
hypertext transfer protocol (HTTP).

Do not be confused by the fact that these two daemons are not on the
same side of a client-server relationship. JDBC allows you to build
systems of Java technology clients (Java clients) that connect to
potentially non-Java programming language servers. Servlets allow
you to build Java technology elements to run on a Web server,
interfacing with potentially non-Java programming language web
clients.

Overview of Distributed Computing 1-13

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
1
Java Database Connectivity (JDBC) Application

The JDBC application encapsulates the specialities of different

database server products. The API provides a set of generic interface
classes that are implemented by a database-specific driver.

This means that the Java 2 SDK alone usually is not sufficient to test a
JDBC application. To connect to a specific database, you need a JDBC
driver from a vendor. However, there is one driver included with the
Java 2 SDK: the JDBC-ODBC bridge driver. When you have a working
open database connectivity (ODBC) driver on a client, you can make a
JDBC connection to the database with only the Java 2 SDK.

Servlets

The Java Servlet API enables you to create Java technology-based

servlets. This API is a standard extension to the Java 2 SDK and is
delivered together with the Java 2 SDK. This means that you can
develop servlets using only the Java 2 SDK. For testing purposes, a
simple servlet runner is available with the Java 2 SDK. However, the
servlet runner only supports HTTP servlets (see the next section) in
single-thread mode.

The most common servlets today are HTTP servlets, which are
modular extensions to a Web server. Common usages of HTTP servlets
are:

● Dynamic hypertext markup language (HTML) page compilation;

that is, getting the contents from other sources (for example, using
the JDBC application from a database)

● Dynamic compilation of a table of contents or an index of an entire

web site

● HTML forms processing, ranging from simple “register me please”

to complex search engines processing query strings

● Dynamic compilation of embedded small parts of HTML pages,

such as visitor counters, clocks, or mini calendars

1-14 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
1
Third-Party Java Technologies Using Daemons
You will find numerous JavaBeans™ components encapsulating
Transmission Control Protocol/Internet Protocol (TCP/IP) protocols,
such as File Transfer Protocol (FTP), Simple Mail Transfer Protocol
(SMTP), Post Office Protocol3 (POP3), Network News Transfer
Protocol (NNTP) and so on. These components save you the effort of
coding the protocol yourself, using Java technology-based sockets.
Also available are APIs or JavaBeans components for non-TCP/IP
legacy protocols. You can even find complete VT100 or 3270 emulators
or emulator Beans on the market.

Understanding Daemons
Discussion – Take a couple of minutes to read through the following
questions, and prepare some answers on your own. Then discuss the
questions in the class.

● List products or technologies you know that are implemented

using daemons.
✓ Most of the TCP/IP standards, like ftp, telnet, yp; but also a lot of “server”-products, like
DB2, Oracle even ORBs use daemons for initial setup.

● If you were to develop a distributed system using daemons, what

would you do?
✓ Decide whether to use an existing protocol or develop a new one; implement a new
daemon or use an existing one; develop an API set for client usage of the daemon;
implement the client using the API set.

● If you have an existing local application and want to distribute it

now, what are the necessary steps to implement it with daemons?
✓ Decide which parts of your application are server, and which are client. Develop an API for
the client so it can use the server. Rewrite the server so that it can function as a daemon.
Rewrite the client so that it uses the API to access the server.

Overview of Distributed Computing 1-15

Overview of Remote Procedure Call

With Remote Procedure Call (RPC), the client calls a local procedure as
if the procedure were in the client’s address space, but the procedure is
actually represented by a proxy and the call is passed to a server.

If you are using a procedural language (such as C or Pascal), RPC

allows you to split up your application onto several distributed
systems without redesigning it too much. Conceptually, local and
remote procedure calls look the same in program code.

It seems likely that RPC will become obsolete because procedural

languages will become obsolete. It is not a good idea to do RPC with
object-oriented programming languages like Java or C++, unless you
have to communicate with a legacy system that happens to offer its
services using RPC.

1-16 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
1
Third-Party Java Technologies Using RPC
The two major legacy RPC systems available today are Sun RPC and
Open Software Foundation/Data Communication Equipment
(OSF/DCE). At the time this course was developed, no Java
technology implementations were available for any of them.

Understanding Remote Procedure Call (RPC)

Discussion – Use a couple of minutes to read through the following
questions, and prepare some answers on your own. Then discuss the
questions in the class.

● List products or technologies you know that are implemented

using RPC.
✓ There are few servers available that are accessed using RPC. RPC is mostly used
internally in distributed applications, which are bought or developed as a whole. One
example is Sun’s NFS environment. A well-known toolkit to support the development of
RPC applications is Open Software Foundation’s Distributed Computing Environment
(OSF/DCE).

● Suppose you have an existing local application written in the Java

programming language that you now want to distribute using
RPC. What are the problems you will encounter?
✓ You have to leave the “thinking model” of your application. Creating instances of objects
and calling methods on them is no longer possible, if they are not on the same machine.
Therefore, the design of the application must be changed at a very fundamental level. As
a guideline, everything remote must be accessible by static methods on one single remote
object.

Overview of Distributed Computing 1-17

Overview of Remote Objects

With remote objects, the client calls methods of a local object. The
objects are represented by proxies, which pass the call request to the
object implementations on other systems.

Therefore, in the remote objects model, programmers deal strictly with

objects: calls to methods are always made through a representative
object. Ideally, programmers treat every object as if it were local. The
actual address space and location of the object are irrelevant. The
programmer deals with the object calls through a set of well-defined
interfaces, and each method call looks like a call through the local
object representation.

In a way, remote objects are to object-oriented languages what RPC is

to procedural languages. The local semantics are extended to
distributed systems as seamlessly as possible.

1-18 Distributed Programming With Java Technology

Implementing Remote Objects With Java Technologies

The Java 2 SDK contains two technologies to do remote objects: the
Java programming language’s native Remote Method Invocation
(RMI) technology and Java Interface Definition Language (JavaIDL)
technology. This section does not compare them in detail. Both
technologies have their strengths and weaknesses. At the end of the
course you should be able to decide which technology to use under
which circumstances.

Remote Method Invocation Technology

RMI was introduced with the JDK™ software, Version 1.1. All you
need to implement and test an RMI application is part of the Java 2
SDK. However, before you put an RMI application into production,
consider carefully your requirements for fault tolerance and
performance. The Java 2 SDK “RMI Registry” (a simple naming
service, as well as a simple implementation repository) is usually too
weak in production environments. As one result of the Enterprise
JavaBeans initiative, robust RMI servers should be available soon.

Overview of Distributed Computing 1-19

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
1
JavaIDL Technology

JavaIDL is part of the Java 2 SDK. The JavaIDL technology allows you
to build and test client and server objects conforming to the CORBA
specification. The JavaIDL technology shares the limitations about
production environments with the RMI technology. To put a JavaIDL
application in production, you should consider using an extremely
robust object request broker (ORB) to host your server objects.

Understanding Remote Objects

Discussion – Use a couple of minutes to read through the following
questions, and prepare some answers on your own. Then discuss the
questions in the class.

● List products or technologies you know that are implemented

using remote objects.
✓ Again, servers that are used via remote objects are not widely used. However, from an
application developer perspective, all the services used in a CORBA environment (naming,
transaction and so on) are accessed using remote objects.

● Remember how you do local event handling in the Java

programming language. Can the same design be applied
remotely? What would be different remotely? Is there a better
solution?
✓ The complete answer to this question is covered in “Object Bus Systems.” A quick
summary is:
• Local event handling is done via a linked list (usually a Vector), holding the references
to the registered event listener objects. The list is processed from top to bottom, and
on each listener object the event handler is called, and the event object passed.
• If the same setup is used for remote object references, partial failure must be taken
into account: If the fifth reference points to a remote object with a currently broken
connection, the event handling essentially stops at this point in time (at least until a
timeout has expired). The rest of the list will get the event after a huge delay.

1-20 Distributed Programming With Java Technology

Overview of Object Bus Systems

With remote objects, you cannot solve the problem of event
distribution. Local event distribution is usually done serially. The
event source goes through a list of registered event listeners and
invokes their event handlers, handing the same event to every listener.
You could do exactly the same process with remote objects; however,
because of the network delays in distributed systems, going through a
large list of event listeners and remotely calling event handlers is slow
and generates a lot of traffic. It also seems unnecessary. The event
being distributed is always the same, so why not post it once on the
wire, and all the interested remote objects can catch the event object as
it passes by?

Essentially an object bus system allows you to post an event to an

abstraction of a wire called a software channel. Listeners interested in
the channel register with the channel so they receive every posting
made to the channel. This frees the event source from having to keep a
list of interested listeners; they have to post only to a specific channel
of registered listeners.

Overview of Distributed Computing 1-21

JavaBeans InfoBus
The JavaBeans Infobus is a type of object bus, but it has a different
focus than “pure” object buses. It facilitates the distribution of data
(rather than the flow of objects) between different JavaBeans objects.
However, because the data to be exchanged is encapsulated in objects,
the InfoBus is still considered to be an object bus. Because the InfoBus
specializes in exchanging data, it supports the following items that are
not supported by a “pure” object bus.

● Semantics of the data on the bus

● Membership protocols

● Security protocols

● Specialized events indicating items, such as “data ready to be

received”

For more information on the InfoBus technology see the JavaBeans

Web site: http://www.java.sun.com/beans/

1-22 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
1
Understanding Object Bus Systems
Discussion – Take a couple of minutes to read through the following
questions, and prepare some answers on your own. Then discuss the
questions in the class.

● Normal TCP/IP connections are point-to-point. If you have the

channel abstraction, but must use normal TCP/IP connections,
what are the consequences?
✓ Basically, you are back to having to send the event to each machine individually. Network
traffic and the latency when sending an event are high. Partial failure, however, can be
handled by the object bus system internally.

● What would be better technologies to use as the transport?

✓ IP broadcast or IP multicast. If there are several different channels, with high volumes of
data each, IP multicast generates a serious load on the clients, because they have to filter
“their” events out. IP multicast delegates this problem to the Ethernet hardware, or at
least to the Ethernet driver level.

Overview of Distributed Computing 1-23

Overview of Mobile Agents

The term mobile agent is used to refer to a variety of items. This section
describes the concept of mobile agents in their broadest sense.

So far, when objects are created they remain on one machine. They use
the network to communicate with each other, using remote method
invocation. While they can exchange other objects with each other
(remember, even a Java programming language String is an object), the
objects being exchanged tend to represent pure data, with accessor
methods to this data.

Mobile agents go one step further: The objects being exchanged carry
behavior as well as data. For example, an object representing a bid to
buy some stock. The “pure data” version contains a desired buying
price and a desired amount. If, at the other end, the price is at your
price or lower, you will make the deal with exactly your price, certainly
not cheaper. Your application creating the bid object is therefore
responsible for getting you a reasonable price.

1-24 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
1
However, the “agent” version can contain a price range and an
amount range, as well as some intelligence to make a deal. It is now
the responsibility of the bid object itself to get you the best deal
possible, while the application creating the bid object merely fills in
your parameters.

Exactly how intelligent these objects are, whether they can travel by
themselves or have to be moved, and whether there is a strong
security focus or no security focus, differentiates the mobile agent
frameworks that are currently available or being researched.
✓ One popular project is the IBM Aglet Workbench. According to the Aglets home page:
“Aglets Workbench is a visual development environment for building aglets, a new breed
of intelligent agent that can travel over a network and execute tasks at the same time.
Aglets combine intelligent agent technology with network-savvy Java technology objects.
They can go from one computer, or Internet host, to another while running and carrying
data with them as they go.”

Overview of Distributed Computing 1-25

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
1
Understanding Mobile Agents
Discussion – Take a couple of minutes to read through the following
question, and prepare some answers on your own. Then discuss the
question in the class.

● Imagine a scenario where a mobile agent could be used.

✓ In many ways, mobile agents are still “a solution seeking a problem to solve”. Systems in
development include network management agents, trading agents, or information-
gathering agents.

1-26 Distributed Programming With Java Technology

Supporting Technologies for Distributed Computing

No matter which technology you decide to use for distributed

computing, some tasks do not change. For example, you almost
always need to locate server machines or objects, knowing nothing
more than a service name. A naming service that resolves this name to
an address comes in handy at this point. There is a whole range of
such supporting technologies. One of the most comprehensive lists is
given in the CORBA specification (http://www.omg.org/). The
following pages describe the most important technologies, including:

● Naming service

● Security service

● Transaction service

● Event service

● Message queueing

Overview of Distributed Computing 1-27

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
1
Naming Service
In a broad sense, this service manages name-to-object associations. You
can bind a name to an object, and you can resolve an object from a
name. A specialized example in use daily is the Domain Name System
(DNS), which resolves domain names, such as java.sun.com to
Internet addresses, 204.160.241.19.

JNDI is intended to be a Java programming language standard

extension, which means it comes free with the Java 2 SDK. The main
goal of JNDI is to provide a standardized API to connect from the Java
programming language to legacy enterprise naming services, such as
lightweight directory access protocol (LDAP), network information
system (NIS), or DNS.

Security Service
Security in large distributed systems, potentially multi-enterprise, is a
difficult issue. For example, objects arriving on a system might claim
to work “on behalf” of some person (remember the “bid” object). How
do you guarantee that this claim is true? How do you prevent the
introduction of malicious code into your system? With the Java 2 SDK,
the Java programming language has most of the necessary basic
mechanisms built in to construct a secure system. However, you still
must construct the security system, which is error prone. A security
service does this for you, giving you a whole system to work with,
instead of several basic, less effective mechanisms.

Transaction Service
A transaction service helps you build transactions. A transaction is a
unit of work that has several characteristics. The most important
characteristic is that a transaction is atomic; this means if a transaction
is interrupted by failure, all effects are undone (rolled back). It also
means that you cannot see intermediate states, such as a temporarily
inconsistent database. Transactions can be local to an address space, on
a remote address space, or span multiple address spaces.

So far, not much is known about the Java Transaction Service. The
following paragraph is cited from the “Java for the Enterprise” home
page:

1-28 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
1
Java Transaction Service (JTS) is a low-level API used by
sophisticated transactional application programs, resource
managers, transaction processing monitors, transaction-aware
communication managers, and transaction managers. Since these
components are provided by different vendors, JTS's role is to
ensure their interoperability in the Java environment.

Event Service
An event service enables you to send and receive events in distributed
systems, in much the same way you do it locally with the regular Java
programming language event mechanism.

Using an event service frees you from thinking about how events are
optimally delivered in distributed systems; you just use the API.
However, when purchasing an event service, be sure to keep the
discussion in Module 6, ‘‘Object Bus Systems,” in mind. How the event
service is implemented has a huge impact on performance and
network traffic.

Message Queueing
A regular RMI call is synchronous. If object A calls object B remotely
using RMI, object A’s calling thread is blocked until the call returns.
This is exactly the same as in a local environment.

However, this is not always what you want. Imagine an application

that lets a user fill out a credit application form. After the data is
collected, an AppForm object is constructed, and, using RMI, a
processAppForm method on the manager’s machine is called. This
might work if the manager is at the desk at this moment and
immediately gives approval. A better solution would be to send the
AppForm object, and start filling out the next application (similar to if
you had sent the application by email). Your client application can
then pick up the processed AppForm object asynchronously, as soon as
it is ready. A message queueing system supports you in building such
an application.

Overview of Distributed Computing 1-29

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
1
Java Message Service

The Java Message Service (JMS) provides a standard API to existing

message oriented middleware. It is the joint work of several enterprise
messaging product vendors.

Supporting Technologies
Discussion – Take a couple of minutes to read through the following
question, and prepare some answers on your own. Then discuss the
question in the class.

● What other supporting technologies do you know or think should

be available?
✓ The CORBA specification lists these CORBA services: Naming, Event Management,
Persistent Object, Lifecycle, Concurrency, Externalization, Relationships, Transactions,
Query, Licensing, Property, Time, Security Services, Trader Object Service, and Collection.

1-30 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
1
Check Your Progress

Before continuing on to the next module, check that you are able to
accomplish or answer the following:

❑ List the five supporting technologies for distributed computing

❑ Compare and contrast the different architectures for distributed

computing

❑ List the distributed computing technologies available for the Java

programming language

Overview of Distributed Computing 1-31

Which of the distributed computing architectures could you use for a

current development project?

1-32 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
Java Database Connectivity (JDBC) 2

Objectives
Upon completion of this module, you should be able to:

● Explain JDBC

● Describe the five major tasks involved in the JDBC programmer’s

interface

● State the requirements of a JDBC driver and its relationship to the

JDBC driver manager

● Explain how to map database types into the Java programming

language

● Describe how to use JDBC with a Java applet

● Compare and contrast two-tier and three-tier designs for JDBC

drivers

● Create a JDBC application to solve a defined problem

The JDBC API enables developers to write code for interfacing with a
database, without knowing the specifics of the database
implementation.

2-1
Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
Relevance

Discussion – Suppose you are tasked with writing an application that

retrieves, updates, modifies, and deletes information from a database.
Consider the following questions:

● What do you need to know about the database?

✓ You do not need to know very much about the database at all. By creating a generic set of
interfaces that can be used to connect to any database vendor’s product, the JDBC API
does not restrict you to a specific database. Moreover, the JDBC API does not restrict you
in the database design, because there are several design alternatives.

● What is the impact on the end-user (client) program if a different

database is used?
✓ The developer writes an application using the interfaces described in the API as though
they were actual class implementations. The driver vendor provides a class
implementation of every interface in the API. If you move your code to a different
database, you need to obtain a different driver implementation for the new database. You
should not have to modify your code.

● What considerations must be made to make the system flexible

and responsive to changes in the business? For example, suppose
that certain customers need to be denied service.
✓ This question gets at the development of three-tiered database systems, where the middle
layer of software abstracts the data interaction from the end-user programs and provides
a means to create business rules, where interactions with the database can be monitored.

Additional Resources

The following resources can provide additional detail on the topics

presented in this module:

● JDBC specification. [Online]. Available:

http://java.sun.com/products/jdbc

● Bowman, Judith S., Sandra L. Emerson, Marcy Darnovsky. 1996.

The Practical SQL Handbook: Using Structured Query Language.
Addison-Wesley.

2-2 Distributed Programming With Java Technology

Introduction

The JDBC API is a set of interfaces designed to insulate a database

application developer from a specific database vendor. The JDBC API
enables the developer to concentrate on writing the application—
making sure that queries to the database are correct and that the data
is manipulated as designed.

With the JDBC, the developer can write an application using the
interface names and methods described in the API, regardless of how
they were implemented in the driver. The developer writes an
application using the interfaces described in the API as though they
are actual class implementations. The driver vendor provides a class
implementation of every interface in the API so that when an interface
method is used, it is actually referring to an object instance of a class
that implemented the interface.

JDBC API also enables developers to pass any string directly to the
driver level. This makes it possible for developers to make use of
custom features of their database without requiring that the
application use only American National Standards Institute (ANSI)
Structured Query Language (SQL).

Java Database Connectivity (JDBC) 2-3

JDBC Drivers and Driver Managers

A JDBC “driver” is a collection of classes that implement the JDBC

interfaces required to connect a Java program to a database. Each
database driver must provide a class that has implemented the
java.sql.Driver interface. The database driver class is used by the
generic java.sql.DriverManager class when it needs a driver to
connect to a particular database using a Uniform Resource Locator
(URL) string. JDBC is patterned after Open Database Connectivity
(ODBC); this makes the task of providing a JDBC implementation on
top of ODBC small and efficient (Figure 2-1 on page 2-6). ODBC is a
standardized API for accessing databases. ODBC provides similar
functionality as JDBC for traditional languages, such as C and C++.

2-4 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
In this class, you use the com.imaginary.sql.msql.MsqlDriver1; a
JDBC driver written to connect to a Mini-SQL database2. The JDBC
implementation is actually written on top of the msql package3—an
API written to connect Java applets and applications to a Mini-SQL
database through a Java socket connection.

1. mSQL-JDBC API is provided courtesy of George Reese.

2. Mini SQL is provided with the courtesy of Hughes Technologies Pty Ltd, Australia
(see http://www.Hughes.com.au).

3. MsqlJava API is provided courtesy of Darryl Collins.

Java Database Connectivity (JDBC) 2-5

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
Figure 2-1 illustrates how a single Java application (or applet) can
access multiple database systems through one or more drivers.

Java application
(JDBC API)

JDBC Driver Manager

URL URL URL URL

JDBC-ODBC
JDBC-NET
driver Driver A Driver B
Bridge driver

ODBC and
JDBC implementation
DB drivers alternatives

Figure 2-1 JDBC Drivers

2-6 Distributed Programming With Java Technology

java.sql Package

There are 18 interfaces associated with the JDBC. They are included in
the following list and shown in Figure 2-2 on page 2-9:

● Driver

● Connection

● Statement

● PreparedStatement

● CallableStatement

● ResultSet

● ResultSetMetaData

● DatabaseMetaData

● ArrayLocator (new in JDBC 2.0)

● Bloblocator (new in JDBC 2.0)

● ClobLocator (new in JDBC 2.0)

Java Database Connectivity (JDBC) 2-7

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
● Ref (new in JDBC 2.0)

● SQLData (new in JDBC 2.0)

● SQLInput (new in JDBC 2.0)

● SQLOutput (new in JDBC 2.0)

● SQLType (new in JDBC 2.0)

● Struct (new in JDBC 2.0)

● StructLoader (new in JDBC 2.0)

✓ All of these interfaces need to be implemented, but whether you do anything with the
methods is up to you.

2-8 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
java.sql Class Hierarchy
java.lang.Object

ArrayLocator
java.lang.Throwable
BlobLocator
java.lang.Exception
ClobLocator
SQLException
StructLocator
BatchUpdateException
SQLData
SQLWarning
Struct
DataTruncation
SQLInput
Connection
SQLOutput
DatabaseMetaData
SQLType
Driver
DriverManager java.util.Date

DriverPropertyInfo Date

Ref Time

ResultSet TimeStamp

ResultSetMetaData Legend
Statement Class

Abstract
PreparedStatement Class
Interface
CallableStatement
Extends
Types
Implements

Figure 2-2 java.sql Class Hierarchy

Java Database Connectivity (JDBC) 2-9

JDBC Flow

Each of the interfaces shown in Figure 2-2 enable you to open

connections to particular databases, execute SQL statements, and
process the results.

● A URL string is passed to the getConnection method of the

DriverManager, which in turn locates a Driver interface.

● With a Driver interface, you can obtain a Connection.

● With the Connection, you can create a Statement.

● When a Statement is executed with an executeQuery method, a

ResultSet can be returned.
✓ A ResultSet object is always returned, but may not contain data when an update or insert
query is executed.

2-10 Distributed Programming With Java Technology

JDBC Programmer’s Interface

This section covers some of the common tasks you perform using the
JDBC programmer’s interface. The following discussion assumes that
you communicate with an mSQL database driver through the JDBC
API. The information is divided into the following parts, which are
separated by lessons:

● Understanding a JDBC example

● Using JDBC drivers (including how to create an instance and

specify a database)

● Opening a database connection

● Using JDBC statements (including how to submit a query and

receive results)

Java Database Connectivity (JDBC) 2-11

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
JDBC Example
The following is a simple example illustrating the common tasks you
perform with JDBC. This example uses the Mini-SQL database in the
lab and shows how to create a Driver instance, a Connection object,
and a Statement object; execute a query; and process the returning
ResultSet object.
1 import java.sql.*;
2
3 public class JDBCExample {
4
5 public static void main (String [] args) throws Exception {
6 if (args.length < 1) {
7 System.err.println("Usage:");
8 System.err.println("JDBCExample jdbc-url");
9 System.exit(1);
10 }
11
12 // Create a JDBC url
13 String url = args[0];
14
15 // The JDBC driver to use
16 String driver = "com.imaginary.sql.msql.MsqlDriver";
17
18 // The query to execute on the database
19 String query = "select * from COFFEES";
20
21 // Load the jdbc driver
22 Class.forName(driver);
23
24 // Use the driver manager to get a connection to the db
25 Connection con = DriverManager.getConnection(url);
26
27 // Use the connection to create a statement
28 Statement stmt = con.createStatement();
29
30 // Execute a query using the statement
31 ResultSet rs = stmt.executeQuery(query);
32
33
34 // Print the result (row by row)
35 while (rs.next()) {
36 System.out.println();
37 System.out.println("Coffee Name: " + rs.getString(1));
38 System.out.println("Supplier Id: " + rs.getInt(2));

2-12 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
39 System.out.println("Price : " + rs.getFloat(3));
40 System.out.println("Sales : " + rs.getInt(4));
41 }
42 }
43 }
44

When the code is run, the contents of the Coffee table in the database
are printed.
java JDBCExample jdbc:msql://<host name>:4333/le-shop

Coffee Name: Colombian

Supplier Id: 101
Price : 7.99
Sales : 0

Coffee Name: French_Roast

Supplier Id: 49
Price : 8.99
Sales : 0

Coffee Name: Espresso

Supplier Id: 150
Price : 9.99
Sales : 0

Coffee Name: Colombian_Decaf

Supplier Id: 101
Price : 8.99
Sales : 0

Java Database Connectivity (JDBC) 2-13

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
Exercise: Compiling a JDBC Application
Exercise objective – Compile and run a simple JDBC application.

Preparation
You need to know the following to complete this exercise:

● How to edit and compile Java technology source code (“Java

source code”) in the environment in which you are working

● The database host location and name

● What JDBC driver you are using and how the URL is called
✓ Assign each student a different database to work with. The databases are named le-shop1
to le-shop9. The URL for the exercise is jdbc:msql://<host name>:4333/<database>,
where <host name> is the machine name designated as the database server and
<database> is replaced with le-shop1 to le-shop9.

Tasks

Compile the JDBC Example

Complete the following steps:

1. Change the directory to labfiles/mod2-jdbc/lab.

2. Compile JDBCExample.java.

Run the JDBC Example

1. Run the example by typing the following in a shell:

java JDBCExample URL

2. Study the results.

Study the Source Code (Optional)

1. Study the source code and get an idea of how JDBC works.

2-14 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
Exercise Summary
Discussion – Take a few minutes to discuss what experiences, issues,
or discoveries you had during the lab exercises.

✓ Manage the discussion here based on the time allowed for this module, which was given
in the “About This Course” module. If you find you do not have time to spend on
discussion, then just highlight the key concepts students should have learned from the
lab exercise.

● Experiences
✓ Ask students what their overall experiences with this exercise have been. You may want to
go over any trouble spots or especially confusing areas at this time.

● Interpretations
✓ Ask students to interpret what they observed during any aspects of this exercise.

● Conclusions
✓ Have students articulate any conclusions they reached as a result of this exercise
experience.

● Applications
✓ Explore with students how they might apply what they learned in this exercise to
situations at their workplace.

Java Database Connectivity (JDBC) 2-15

Using JDBC Drivers

There are two ways to create a JDBC Driver instance: explicitly, or
through the jdbc.drivers property.

Explicitly Creating an Instance of a JDBC Driver

To communicate with a particular database engine using JDBC, you

must first create an instance of the JDBC driver. The driver has a static
initializer that registers an instance of itself with the driver manager.
This driver remains behind the scenes, handling any requests of that
type of database.
// Load the driver class
Class.forName("com.imaginary.sql.msql.MsqlDriver");

2-16 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
To load and register a driver, you can also create an instance of the
driver yourself.
// Create an instance of the JDBC Driver by Imaginary
new com.imaginary.sql.msql.MsqlDriver();

It is not necessary to associate this driver with a variable; the driver

exists after it is instantiated and successfully loaded into memory.
✓ The first method is the preferred way to load and register a driver.
✓ No variable is needed with new com.imaginary.sql.msql.MsqlDriver because the
constructor creates a static instance of itself that remains in memory because of a system
reference. Also note that the driver is responsible for calling the
DriverManager.registerDriver(this) method.

Java Database Connectivity (JDBC) 2-17

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
Exercise: Loading a Driver

Exercise objective – Learn how to register a driver with the driver

manager.

Tasks

Write a Program to Load and Register a Driver With the Driver

Manager

Complete the following steps:

1. Change the directory to labfiles/mod2-jdbc/lab.

2. Write a simple class LoadDriver, which loads the mSQL driver.

Use the methods defined by the Driver interface to print
information, such as the driver’s name and version on the console.
Use LoadDriver.java as a template.

3. Compile and run your application. Use the same JDBC URL as in
the previous exercise.

Experiment With the Driver

1. The LoadDriver program never accesses any data in the database,

so try to shorten the provided JDBC URL until the program does
not work anymore. What is the shortest URL?
✓ The minimal URL for the mSQL driver is jdbc:msql:. The hostname, table, username and
password are not needed by the driver manager to find the mSQL driver.

2. Find out why the mSQL driver is not JDBC compliant (optional).
✓ The mSQL database does not implement all features needed for JDBC compliance.
Therefore the driver is not compliant either.

2-18 Distributed Programming With Java Technology

● Experiences
✓ Ask students what their overall experiences with this exercise have been. You may want to
go over any trouble spots or especially confusing areas at this time.

● Interpretations
✓ Ask students to interpret what they observed during any aspects of this exercise.

● Conclusions
✓ Have students articulate any conclusions they reached as a result of this exercise
experience.

● Applications
✓ Explore with students how they might apply what they learned in this exercise to
situations at their workplace.

Java Database Connectivity (JDBC) 2-19

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
Loading JDBC Drivers Through jdbc.drivers

You can have more than one database driver loaded into memory. You
can also have more than one of the drivers loaded into memory, ODBC
or a JDBC generic network protocol is capable of connecting to the
same database. If this is the case, JDBC allows you to specify a list of
drivers in a specified order. The order of selection is specified by a Java
programming language properties tag, jdbc.drivers. The
jdbc.drivers property should be defined as a colon-separated list of
driver class names:
jdbc.drivers = com.imaginary.sql.msql.MsqlDriver:Acme.cool.driver

Properties are set through the -D option to the java interpreter (or the
-J option to the appletviewer application). For example:
java -Djdbc.drivers=com.imaginary.sql.msql.MsqlDriver:Acme.cool.driver

When attempting to connect to a database, JDBC uses the first driver it

finds that can successfully connect to the given URL. It first tries each
driver specified in the properties list, in order from left to right. It then
tries any drivers that are already loaded in memory, in the order that
the drivers were loaded. If the driver was loaded by untrusted code, it
is skipped unless it has been loaded from the same source as the code
that is trying to open the connection.

2-20 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
Registering a Driver

In either case, when a driver is loaded, it is the responsibility of the

driver implementation to register itself with the driver manager. For
example, the Mini-SQL driver, com.imaginary.sql.MsqlDriver
constructor looks like the following:
1 public class MsqlDriver implements java.sql.Driver {
2 static {
3 try {
4 new MsqlDriver();
5 }
6 catch(SQLException e) {
7 e.printStackTrace();
8 }
9 }
10 /**
11 * Constructs a new driver and registers it with
12 * java.sql.DriverManager.registerDriver() as specified by the JDBC
13 * draft protocol.
14 */
15 public MsqlDriver() throws SQLException {
16 java.sql.DriverManager.registerDriver(this);
17 }
18 ...
19 }

The JDBC driver by Imaginary creates an instance of itself in the static

code block (lines 2–9).

When the constructor is called, either explicitly or implicitly, the driver

registers itself with the driver manager (line 16).
✓ Note that as soon as the class is loaded, an instance of the driver is registered with the
driver manager. Therefore use Class.forName(<class name>) to register the driver.

Java Database Connectivity (JDBC) 2-21

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
Exercise: Loading Driver 2 (Optional)

Exercise objective - Learn how JDBC drivers are registered implicitly

with the driver manager.

Tasks

Remove the Explicit Driver Registration From LoadDriver

Complete the following steps:

1. Change the directory to labfiles/mod2-jdbc/lab.

2. Remove the explicit driver registration from your

LoadDriver.java (or use the already complete
LoadDriver2.java).

3. Compile and run the program. Set the name of the drivers class
(com.imaginary.sql.msql.MsqlDriver) in the jdbc.drivers
property.
java -Djdbc.drivers=com.imaginary.sql.msql.MsqlDriver
LoadDriver2 jdbc:msql://<host name>:4333/le-shop
✓ Note – The property jdbc.drivers does not determine the sequence in which JDBC
drivers are checked. It eliminates the need for any class.forName(“...”) statement.

2-22 Distributed Programming With Java Technology

● Experiences
✓ Ask students what their overall experiences with this exercise have been. You may want to
go over any trouble spots or especially confusing areas at this time.

● Interpretations
✓ Ask students to interpret what they observed during any aspects of this exercise.

● Conclusions
✓ Have students articulate any conclusions they reached as a result of this exercise
experience.

● Applications
✓ Explore with students how they might apply what they learned in this exercise to
situations at their workplace.

Java Database Connectivity (JDBC) 2-23

Specifying a Database
Now that you have created the instance of the JDBC driver, you need
to name the database to which you want to connect.

In JDBC, you do this by specifying a URL that indicates the database

type. The proposed URL string syntax for a JDBC database is the
following:
✓ Note – This is a URL-like string, not a Java URL object.

jdbc:subprotocol:parameters

where subprotocol names a particular kind of database connectivity

mechanism that can be supported by one or more drivers. The
contents and syntax of the parameters depends on the subprotocol.
// Construct the URL for JDBC access
String url = args[0];

This is the URL for the JDBC access to the mSQL Ticketing database
you will connect to in the lab. serverName is a variable set to the host
name of the database server.

2-24 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
The subprotocol in the previous instance is msql. It could have been
any other type of protocol that is accessed through a JDBC-ODBC
bridge; for example:
jdbc:odbc:Object.le-shop

Java Database Connectivity (JDBC) 2-25

NIS Name Resolution Example

You can use a subprotocol, such as NIS to resolve the name of a
specified database:
jdbc:nisnaming:le-shop

The JDBC URL mechanism is intended to provide a framework so that

different drivers can use various naming systems that are appropriate
to their needs. Each driver only has to understand one URL naming
syntax to enable it to reject other types of URLs, thus providing
another layer of security.

2-26 Distributed Programming With Java Technology

Opening a Database Connection

Now that you have created a URL specifying msql as the database
engine, you are ready to make a database connection.

To do this, obtain a java.sql.Connection object by calling the JDBC

driver’s java.sql.DriverManager.getConnection method.
con = DriverManager.getConnection(url);

The following describes the process:

● The driver manager calls the Driver.getConnection method of

every driver registered, passing the URL string as the parameter.

● If the driver “knows” the subprotocol name, then the driver

returns an instance of a Connection object; otherwise it returns
null.

Java Database Connectivity (JDBC) 2-27

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
Figure 2-3 illustrates how a driver manager resolves a URL string
passed in the getConnection method. When the driver returns a null,
the driver manager continues to call the next registered driver in turn
until either the list is exhausted or a Connection object is returned.

URL string
Driver manager Program

getConnection(URL string);

jdbc:A jdbc:B jdbc:msql

driver driver driver

Explanation Connection to
le-shop DB
Driver manager calls getConnection(URL),
which calls driver.connection(URL) for the
drivers in the vector until a match is found.

The URL is parsed (jdbc:drivername).

When the driver in the vector matches the parsed le-shop

drivername, a connection to the database is
made.

If the driver does not match, NULL is returned and

the next driver in the vector is checked.

Figure 2-3 Example of Database Resolution

✓ Note that the driver does not have a “connection” with the database—this is the job of the
Connection interface implementation.

2-28 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
Exercise: Connecting to a Database

Exercise objective – Learn how to connect to a given database and

how to gather meta data.

Tasks

Write a Simple Program to Connect to a Database

Complete the following steps:

1. Change the directory to labfiles/mod2-jdbc/lab.

2. Write a simple class FirstConnection, which loads the mSQL

driver and connects to the given mSQL database. Use
FirstConnection.java as a template.

3. Use the getMetaData method and the DatabaseMetaData class to

gather information about the connection. Print the database name
and version number on the console.

4. Compile and run the program.

Study How the Tables Are Listed (Optional)

1. The last lines of FirstConnection.java shows how the tables in

a database are listed. Study the code and get an idea of how the
ResultSet class works.
✓ The getTables method uses a parameter to narrow the list of tables to certain types.
mSQL does not support any of the possible parameters.

Java Database Connectivity (JDBC) 2-29

● Experiences
✓ Ask students what their overall experiences with this exercise have been. You may want to
go over any trouble spots or especially confusing areas at this time.

● Interpretations
✓ Ask students to interpret what they observed during any aspects of this exercise.

● Conclusions
✓ Have students articulate any conclusions they reached as a result of this exercise
experience.

● Applications
✓ Explore with students how they might apply what they learned in this exercise to
situations at their workplace.

2-30 Distributed Programming With Java Technology

Submitting a Query
To submit a standard query, get a Statement object from the
Connection.createStatement method.
try {
stmt = con.createStatement();
} catch (SQLException e) {
System.out.println (e.getMessage());
}
✓ SQL exceptions occur when there is a database access error; that is, a connection is
broken or the database server went down. SQL exceptions provide information for
debugging, such as
• A string describing the error
• An SQL state string following XopenSQL state conventions
• A vendor-specific integer error code

Java Database Connectivity (JDBC) 2-31

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
Use the Statement.executeQuery method to submit the SQL
statement to the database. JDBC passes the SQL statement to the
underlying database connection unaltered. JDBC does not attempt to
interpret queries.
ResultSet rs = stmt.executeQuery("select *
from COFFEES order by SALES");

The Statement.executeQuery method returns a ResultSet for

processing.

2-32 Distributed Programming With Java Technology

Using a Prepared Statement

If the same SQL statements are going to be executed multiple times,
you should use a PreparedStatement object. A prepared statement is a
precompiled SQL statement that is more efficient than calling the same
SQL statement over and over. The PreparedStatement class extends
the Statement class to add the capability of setting parameters inside
of a statement. An example of using a prepared statement is shown in
the following code.
✓ Pre-compilation of multiple statements occurs at the database. The database (if it
supports it) can take the entire command string and execute it against the database in a
single execution and then return the results. The Imaginary JDBC driver for mSQL does
not support pre-compiled statements, so each one is executed as an individual
transaction.

Java Database Connectivity (JDBC) 2-33

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
Note – Index numbers in JDBC start from 1, not 0.

1 public boolean prepStatement(float sales, String name){

2 PreparedStatement prepStmnt = con.prepareStatement(
3 "update COFFEES set SALES = ? where COF_NAME = ?");
4 prepStmnt.setFloat(1, sales);
5 prepStmnt.setString(2, name);
6 int rowsUpdated = prepStmnt.executeUpdate();
7 return rowsUpdated > 0;
8 }

The setXXX methods in Table 2-1 on page 2-34 for setting SQL IN
parameter values must specify types that are compatible with the
defined SQL type of the input parameter. For example, if the IN
parameter has SQL type Integer, then you should use setInt.
✓ IN parameters are passed by value into the operation; the value of the parameter is not
expected to be changed.
✓ OUT parameters are passed by reference from the operation; the operation is to set the
value of the reference.
✓ INOUT parameters are passed by reference into an operation; the value of the parameter
passed into the operation is expected to be changed by the operation.
✓ JDBC only supports IN and OUT, but JavaIDL (CORBA) supports all three.

The setXXX Methods

Table 2-1 contains the setXXX methods and SQL types.

Table 2-1 setXXX Methods and SQL Types

Method SQL Type(s)

setArrayLocator LOCATOR(<array>)
setASCIIStream Uses an American Standard Code for
Information Exchange (ASCII) stream to
produce a LONGVARCHAR
setBigDecimal NUMERIC
setBinaryStream Uses a binary stream to produce a
LONGVARBINARY
setBlobLocator LOCATOR(BLOB)
setBoolean BIT

2-34 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
Table 2-1 setXXX Methods and SQL Types

Method SQL Type(s)

setByte TINYINT
setBytes VARBINARY or LONGVARBINARY (depending upon
the size relative to the limits on VARBINARY)
setCharacterStream Uses a java.io.Reader to produce a
LONGVARCHAR
setClobLocator LOCATOR(CLOB)
setDate DATE
setDouble DOUBLE
setFloat FLOAT
setInt INTEGER
setLong BIGINT
setNull NULL
setObject The given Java technology object (“Java object”)
is converted to the target SQL Type before being
sent.
setShort SMALLINT
setString VARCHAR or LONGVARCHAR (depending upon the
size relative to the driver’s limits on VARCHAR)
setStructLocator LOCATOR(<structured-type>)
setTime TIME
setTimestamp TIMESTAMP
setUnicodeStream Uses a Unicode stream to produce a
LONGVARCHAR

Java Database Connectivity (JDBC) 2-35

Using Callable Statements

A callable statement allows non-SQL statements (such as stored
procedures) to be executed against the database. The
CallableStatement class extends the PreparedStatement class,
which provides the methods for setting IN parameters. Because the
PreparedStatement class extends the Statement class, methods for
retrieving multiple results with a stored procedure are supported with
the Statement.getMoreResults method.

For example, you could use a CallableStatement if you wanted to

store a precompiled SQL statement to query a database containing the
coffee inventory or suppliers.
1 String coffeeName= "Espresso";
2 CallableStatement querySales = con.prepareCall("{call
return_sales[?, ?]}");
3 try {
4 querySales.setString(1, coffeeName);
5 querySales.registerOutParameter(2, java.sql.Type.REAL);
6 querySales.execute();
7 float sales = querySales.getFloat(2);
8 } catch (SQLException e){

2-36 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
9 System.out.println("Query failed");
10 e.printStackTrace();
11 }

Before executing a stored procedure call, you must explicitly call

registerOutParameter to register the java.sql.Type of any SQL
OUT parameters.

Java Database Connectivity (JDBC) 2-37

Receiving Results
The result of executing a statement can be a table of data that is
accessible using a java.sql.ResultSet object. The table consists of a
series of rows and columns. The rows are received in order. A
ResultSet keeps a cursor pointing to the current row of data and is
initially positioned before its first row. The first call to next makes the
first row the current row, the second call makes the second row the
current row, and so forth.

The ResultSet object provides a set of get methods that enable access
to the various column values of the current row. These values can be
retrieved using either a column name or an index. You should use an
index when referencing a column. Column indexes start at 1. When
using a name to reference a column, it is possible that more than one
column will have the same name, thus causing a conflict.

2-38 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
Note – The name used to reference a column is case sensitive.

1 while (rs.next()) {
2 System.out.println();
3 System.out.println("Coffee Name: " + rs.getString(1));
4 System.out.println("Supplier Id: " + rs.getInt(2));
5 System.out.println("Price : " + rs.getFloat(3));
6 System.out.println("Sales : " + rs.getInt(4));
7 }

The various getXXX methods access columns within the result sets
table. The columns can be accessed in a random order, within the
specified row.

To retrieve data from the ResultSet object, you must be familiar with
the columns returned and their data types. A table mapping Java
technology types (“Java types”) to SQL data types is provided in
Table 2-3 on page 2-43.

Note – Once you have read the ResultSet object, the results are
cleared; you can read the results only once.

Java Database Connectivity (JDBC) 2-39

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
Exercise: Executing an SQL Query

Exercise objective – Learn how to send SQL queries to the database

and how to process the results.

Tasks

Write a Program to Send a SQL Query to a Database

Using the template Exec.java,complete the run method. The query

is stored in queryString, send it to the database and display the
result on the console. Complete the following steps:

1. Change the directory to labfiles/mod2-jdbc/lab.

2. Get a Statement object from the connection.

3. Send the query stored in queryString to the database. Store the

results in a ResultSet object.

4. Process the ResultSet and display the result on the console.

(getString(...) can be used for every member in the column.)

5. Compile and run the program. Possible queries are:

▼ select * from COFFEES

▼ select * from SUPPLIERS

▼ select COF_NAME from COFFEES where SUP_ID=101

Experiment Further (Optional)

1. Use metadata to add the column’s label to the output.

2. Use SQL queries that do not return ResultSet objects, such as:

▼ insert into COFFEES values(’My Coffee’,150,5.55,0)

3. Did you use executeQuery or executeUpdate to execute the

query? Why?

2-40 Distributed Programming With Java Technology

● Experiences
✓ Ask students what their overall experiences with this exercise have been. You may want to
go over any trouble spots or especially confusing areas at this time.

● Interpretations
✓ Ask students to interpret what they observed during any aspects of this exercise.

● Conclusions
✓ Have students articulate any conclusions they reached as a result of this exercise
experience.

● Applications
✓ Explore with students how they might apply what they learned in this exercise to
situations at their workplace.

Java Database Connectivity (JDBC) 2-41

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
Using JDBC Statements
Table 2-2 contains the getXXX methods and the returned Java type.

Table 2-2 getXXX Methods and the Java Type Returned

Method Java Type Returned

getArrayLocator LOCATOR(<array>)
getASCIIStream java.io.InputStream
getBigDecimal java.math.BigDecimal
getBinaryStream java.io.InputStream
getBlobLocator LOCATOR(BLOB)
getBoolean boolean
getByte byte
getBytes byte[]
getCharacterStream java.io.Reader
getClobLocator LOCATOR(CLOB)
getDate java.sql.Date
getDouble double
getFloat float
getInt int
getLong long
getObject Object
getShort short
getString java.lang.String
getStructLocator LOCATOR(<structured-type>)
getTime java.sql.Time
getTimestamp java.sql.Timestamp
getUnicodeStream java.io.InputStream or Unicode characters

✓ Obviously, getNull makes no sense, but if the getXXX method encounters an SQL NULL
type, the result of the method is null.

2-42 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
Mapping SQL Types to Java Types

Table 2-3 shows the standard Java types for mapping various common
SQL types.

Table 2-3 Mapping SQL Types to Java Types

SQL Type Java Type

CHAR String
VARCHAR String
LONGVARCHAR String
NUMERIC java.math.BigDecimal
DECIMAL java.math.BigDecimal
BIT boolean
TINYINT byte
SMALLINT short
INTEGER int
BIGINT long
REAL float
FLOAT double
DOUBLE double
BINARY byte[]
VARBINARY byte[]
LONGVARBINARY byte[]
DATE java.sql.Date
TIME java.sql.Time
TIMESTAMP java.sql.Timestamp

Java Database Connectivity (JDBC) 2-43

JDBC Driver Architecture

By allowing you to create a generic set of interfaces that you can use to
connect to any database vendor’s product, the JDBC API does not
restrict you to a specific database. In addition, the JDBC API also does
not restrict you in the application design, because you use either a
two- or three-tier model to build your application.

JDBC Driver Categories

Today’s JDBC drivers generally fit into one of four categories:

1. JDBC-ODBC bridge plus ODBC driver. A bridge provides JDBC

access to the database using an existing ODBC driver. The ODBC
driver uses native code, which must be installed on each database
client. Therefore, this driver is most appropriate for corporate
networks where client installations are not a major problem, or for
an application server in a three-tier architecture.

2-44 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
2. Native-API partly-Java technology driver (“Java driver”). This kind of
driver is built on top of a native database client library (usually C
or C++). The driver translates the JDBC calls into calls to the API
of the database client library. The library is distributed by the
provider of the database management system (DBMS). This driver
is similar to the ODBC bridge, because it needs some native code
installed on each database client.

3. JDBC-Net pure Java driver. This kind of driver communicates with

an intermediary server seated between the client and the database,
using a network protocol specific to the intermediary. These calls
are translated by the intermediary to DBMS-specific calls and
forwarded to the database. No native code needs to be installed on
the database client machines, because the driver can be
implemented in pure Java technology. If the intermediary can talk
to different DBMSs, you need only one JDBC driver.

4. Native-protocol pure Java driver. This kind of driver translates JDBC

calls directly into the network protocol used by the DBMS. This
allows direct calls from the client to the database. Because these
protocols are proprietary, the database provider is the primary
source for the driver. Several vendors have these in progress.

Drivers from categories 3 and 4 are the preferred way to access

databases from JDBC, because they are pure Java technology. Drivers
from categories 1 and 2 are considered as interim solutions until pure
Java drivers are available. Table 2-4 shows the four categories and their
properties.

Table 2-4 Driver Categories

Driver Category Pure Java Net Protocol

JDBC-ODBC Bridge No Direct

Native API based No Direct
JDBC-Net Yes Connector
Native Protocol Yes Direct

Java Database Connectivity (JDBC) 2-45

Application Designs

By creating a generic set of interfaces that can be used to connect to

any database vendor’s product, the JDBC API does not restrict you to
a specific database. In addition, the JDBC API does not enforce a
special application design. This section discusses two- and three-tier
models as guidelines to database application design with the Java
programming language and the JDBC program. An application
consists of a database client, a database server, and some business or
application logic.

Two-Tier Application Designs

An application designed after the two-tier model consists of a client as
the top tier and a database server as the bottom tier. The client usually
has a graphical user interface (GUI) to interact with the user and
includes all the business or application logic. Furthermore, the client
talks directly to the database using an appropriate JDBC driver. This
design is straightforward and intuitive.

2-46 Distributed Programming With Java Technology

Three-Tier Application Designs

An application designed after the three-tier model has a middle tier
between the client and the database. Within this model, the client is
much thinner and essentially consists of the GUI. Most of the business
or application logic is moved to the middle tier. The middle tier offers
a more abstract API to the client and they use any protocol, such as
RMI, or CORBA to communicate. The middle tier uses JDBC to
communicate with the database server, which plays the same role as in
the two-tier model.

The three-tier model is the preferred model to design database

applications. The added level of abstraction eases development and
maintenance of the whole application. Furthermore, it is easier to
integrate several databases or even legacy systems in an application.

Java Database Connectivity (JDBC) 2-47

Applets

The use of Java applets has been the most publicized use of the Java
platform. JDBC can be incorporated into applets to bring database
access to the World Wide Web (WWW). For example, you can
download a Java applet that can display the available flights for a
specified date to and from specified destinations. This applet could
access a relational database over the Internet, enabling a client to
inquire about seat availability, book a reservation, or update the
database.

You can also use applets in an intranet to provide access to company

databases, like a corporate directory, where many divisions are
working on different hardware platforms, but need a common
database interface.

2-48 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
Applets and Traditional Database Applications
Applets differ from traditional database applications in a number of
ways:

● Untrusted applets are severely constrained in the operations they

are allowed to perform. Specifically, untrusted applets are
prevented from having access to local files and the ability to open
a network connection to any host other than the host that
provided the applet.

● Applets running across the Internet cannot depend on a local

database location or the database driver being in a .INI file or
local registry on the client’s machine, as in ODBC.

Performance Considerations
Performance considerations for database connectivity implementations
differ when the database might be halfway around the world. The
response time for an Internet-based database applet will be
considerably slower than a database applet running on a local
network.

Security Limitations
You can avoid some of the security limitations encountered with
untrusted applets by using a digital signature or cryptographic key
scenario. In these circumstances, an applet is treated like an
application in the security sense, but there still would be problems
interacting with client-side databases because of the difficulty locating
the directory structure of the database or database driver.

As discussed earlier, a three-tier database access design can provide a

middle tier of service on the network. The middle tier
implementations can access databases on multiple networked hosts.
These calls might be made through remote procedure call (RPC) or
through an object request broker (ORB). In either case, the middle tier
is best defined using an object paradigm; for example, customer
objects with operations for customer invoicing, booking reservations,
and other transactions.

Java Database Connectivity (JDBC) 2-49

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
Exercise: Building a JDBC Application (Optional)

Exercise objective – Combine all of the aspects shown earlier and use
JDBC to build a larger application.

Preparation
LeShop is a small coffee shop that uses a database (le-shop) for its
business. The example’s focus is on JDBC, so there is a minimum of
supporting functions, no GUI, and a simple two-tier design. It consists
of five classes: LeShop is the main application. It uses ConsoleMenu
and ConsoleDialog for user interaction on the console. The classes
Coffee and Supplier interact with the database using JDBC. They
both have a main method, so you can use them as standalone
programs to test the code. Use FirstConnection and Exec from the
previous exercises to verify that your code manipulates the database
correctly.

Tasks

Set up the LeShop Tables

The classes Coffee and Supplier offer methods to create, initialize,

and remove the necessary tables.

1. Change the directory to labfiles/mod2-jdbc/lab.

2. Complete the insert(...) method in Coffee.java. Compile and

test it. Use FirstConnection or Exec from previous exercises to
verify the results.

Did you use executeQuery or executeUpdate to execute the

query? Why?
✓ ExecuteUpdate would be the correct method to use, because neither insert nor drop table
return a ResultSet. Of course, executeQuery works too.

3. Complete the removeTables method in Coffee.java. Use Exec

to test your SQL queries prior to coding them.

2-50 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
4. Compile and test the removeTables method in Coffee.java. Use
FirstConnection or Exec from previous exercises to verify the
results.

5. Repeat steps 2–5 using the Supplier class. (Optional – If the

Supplier class is left out, copy the Supplier1.java from the
solutions directory and rename it to Supplier.java.)

6. Compile LeShop.java and test the functioning parts.

Complete Sales Total

Complete the getSalesTotal methods in Supplier and Coffee and

run LeShop again. The application is almost complete now.

1. Complete getSalesTotal in Coffee.java.

2. Compile Coffee.java and run LeShop to test the code.

3. Repeat steps 7–8 for Supplier.java.

Delete Single Entries (Optional)

Implement the delete(...) methods from Coffee and Supplier.

The application will be complete afterwards.

1. Complete the delete(...) method in Coffee.java. You can use

Exec to test your SQL queries prior to coding them.

2. Compile Coffee.java and run LeShop to test the code.

3. Repeat steps 10–11 for Supplier.java.

Tasks

Have Further Discussion (Optional)

The initTable method from Supplier and Coffee could be

much better designed using a PreparedStatement. What would
the code look like?

Beside the GUI, would it be difficult to turn the LeShop

application into an applet?

Java Database Connectivity (JDBC) 2-51

● Experiences
✓ Ask students what their overall experiences with this exercise have been. You may want to
go over any trouble spots or especially confusing areas at this time.

● Interpretations
✓ Ask students to interpret what they observed during any aspects of this exercise.

● Conclusions
✓ Have students articulate any conclusions they reached as a result of this exercise
experience.

● Applications
✓ Explore with students how they might apply what they learned in this exercise to
situations at their workplace.

2-52 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
More Information

For more information on JDBC, refer to the latest revision of the JDBC
specification. This specification is available at the
http://java.sun.com/products/jdbc web site. It specifies the
interface both from the points of view of the application programmer
and the writer of the vendor driver.

Java Database Connectivity (JDBC) 2-53

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
2
Check Your Progress

Before continuing on to the next module, check that you were able to
accomplish the following in this module:

❑ Explain JDBC

❑ Describe the five major tasks involved in the JDBC programmer’s

interface

❑ State the requirements of a JDBC driver and its relationship to the

JDBC driver manager

❑ Explain how to map database types into the Java programming

language

❑ Describe how to use JDBC with a Java applet

❑ Compare and contrast two-tier and three-tier designs for JDBC

drivers

❑ Create a JDBC application to solve a defined problem

2-54 Distributed Programming With Java Technology

To what applications will you add JDBC?

Java Database Connectivity (JDBC) 2-55

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
Remote Method Invocation (RMI) 3

Objectives
Upon completion of this module, you should be able to:

● Describe the RMI architecture, its layers and garbage collection

● Implement an RMI server and client in the Java programming

language

● Generate client stubs and skeletons for remote services using the
RMI stub compiler

● Define the RMI registry and describe how it works

● Explain the security issues related to RMI

The Java Remote Method Invocation (RMI) is a simple and powerful

framework for distributed object computing that extends the pure Java
object model to the network.

3-1
Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
Relevance

Discussion – Consider the following questions:

● How many of the programs that you write could take advantage
of existing programs/classes that you have already written in the
Java programming language, but need to run on another machine
elsewhere on the network?
✓ The RMI API may be viewed as a communication mechanism that takes care of TCP socket
connectivity and putting objects “on the wire” (using the Java programming language’s
Object Serialization) between separate JVM invocations. The developer is removed one
level by not having to write that functionality into the program each time.

● As you survey the distributed programming technologies

available to you, how important is it that the chosen technology
supports the ability to pass Java objects “across the wire” as
method parameters and/or return values?
✓ It should be very important to anyone who has taken Java Programming Workshop (SL-285), or
anyone who had to pass complex data structures from one machine to another before the
Java programming language supported object serialization. At that time, you had to write
your own protocol to pass the data states across the wire, and a protocol handler to
receive the data and instantiate a new object of a particular type, once the data has been
received.
✓ Keep in mind that CORBA does not allow objects to be passed across the wire, just
primitive types, predefined composite types, and CORBA references.

3-2 Distributed Programming With Java Technology

● How important is it that the distributed technology that you

implement supports the ability to pass complex data types, for
example, a variable-size Vector of Java objects? How about the
capability to pass or return a subclass or subtype of the declared
parameter or return value?
✓ Again, this is an inherent advantage to using RMI over CORBA, because CORBA returns
must be strictly predefined in size and type, and are limited to primitives. The subclass,
though, is incredibly powerful, in that you could declare your method to accept a Command1
object. You could then subclass Command into Deposit and Withdraw, and RMI would allow
you to pass either Deposit or Withdraw to your method. CORBA does not allow you the
flexibility of passing subtypes.
✓ Once you understand this, imagine that the Command object implements runnable, so from
your method, you can invoke the run method of the object that you have received. Its
behavior will be different, depending on whether it is a Deposit or a Withdraw object. So
you are passing behavior rather than just data, which is the process of implementing an
agent.

1. See “Command pattern,” pp. 233–242 of Design Patterns – Elements of Reusable Object-Oriented
Software by Gamma, Helm, Johnson, and Vlissedes (Addison-Wesley, 1995).

Remote Method Invocation (RMI) 3-3

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
Additional Resources

Additional resources – The following resources can provide

additional details on the topics presented in this module.

● The Java Remote Method Invocation Specification. [Online].

Available: http://java.sun.com/products/jdk/1.2/docs/
guide/rmi/spec/rmiTOC.doc.html

● Getting started using RMI. [Online]. Available:

http://java.sun.com/products/jdk/1.2/docs/guide/
rmi/getstart.doc.html

● Frequently Asked Questions, RMI and Object Serialization.

[Online]. Available: http://java.sun.com/products/jdk/
rmi/faq.html

● Java Remote Method Invocation – Distributed Computing for Java.

[Online]. Available: http://java.sun.com/marketing/
collateral/javarmi.html

● Object Serialization. [Online]. Available: http://java.sun.com/

products/jdk/rmi/serial/index.html

3-4 Distributed Programming With Java Technology

What Is Java RMI?

Java Remote Method Invocation (RMI) is the method preferred by Sun
Microsystems to write distributed objects using the Java environment.
RMI extends the Java slogan “Write Once, Run Anywhere”™ with
“Run Everywhere.” The distributed object model is in fact seamlessly
integrated into the Java programming language.

Until the release of the Java Remote Method Invocation API, sockets
were the only facility built into the Java programming language that
provided direct communication between machines. In a fashion
similar to Remote Procedure Calls (RPC), RMI abstracts the socket
connectivity and data streaming, enabling developers to write code
that accesses remote objects in the same fashion as objects instantiated
from within the local runtime system.

RMI supports traditional distribution designs, such as client/server or

peer-to-peer, and enriches these models with new kinds of agent-based
distributed applications that move behavior between clients and
servers.

Remote Method Invocation (RMI) 3-5

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
RMI enables calls to remote objects that exist in the runtime of a
different Java Virtual Machine (JVM) invocation. This “remote” or
“server” JVM can be executing on the same machine or on an entirely
different machine from the RMI “client.” A Java program can make a
call on a remote object once it obtains a reference to the object, either
by looking it up in the registry (a simple naming facility provided by
RMI) or by receiving the reference as an argument or return value.

Unlike CORBA, your application must be written entirely in the Java

programming language and is therefore a pure Java-to-Java
mechanism providing a consistent environment for distributed
applications. RMI is in fact an extension to the language itself, and was
designed to ease the design and implementation of distributed
applications, at the price of staying within the Java environment. No
separate interface definition language (IDL) or special language-
neutral environment is used. Everything has to be written in the Java
programming language.

3-6 Distributed Programming With Java Technology

RMI Characteristics
The characteristics of RMI include the following:

● RMI is object-oriented. It passes objects as arguments or return

values and not some predefined data types. It allows you to
handle even the most complex objects as a single argument. No
composing or decomposing into primitive data types is needed.

● RMI is a simple but powerful framework. It allows you to build

the core part of server and client applications with just a few lines
of code.

● RMI benefits directly from the platform independence of the Java

programming language. Any RMI-based application is portable to
any Java virtual machine.

● RMI allows you to move behavior, such as agents or business logic

code to the part of your network where it makes the most sense
for your application. With the ability to pass behavior comes the
ability to directly implement object-oriented design patterns that
rely on different behavior schemes.

Remote Method Invocation (RMI) 3-7

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
● The Java programming language provides built-in security
mechanisms that are also used by RMI, such as a security manager
protecting the system from hostile code.

● RMI has built-in distributed garbage collection that removes

remote server objects that are no longer referenced by any clients
in the network.

● RMI is inherently multithreaded and therefore allows you to create

sophisticated servers using concurrent threads to process client
requests.

● RMI also allows a connection to existing server systems using the

Java native interface (JNI) or to standard relational databases with
the JDBC classes.

3-8 Distributed Programming With Java Technology

RMI Application Architecture

✓ Do not spend too much time on this overhead. Everything on it will be covered in detail
later on. The purpose of this overhead is to give an overview, which should not take more
than five minutes.

RMI allows you to move behavior from server to client and from client
to server. Imagine an application for an insurance company allowing
you to calculate some installments. The interface for this application
certainly includes some methods that check whether the entered
figures conform to values that are set by the company rules. To fill out
an insurance form, clients get an object from the server that
implements these rules. When these rules change, clients get the
updated implementation the next time they request the insurance form
object. The clients’ interface does not need to be touched at all. The
checking takes place at the client side with the advantage of
immediate feedback to the user and reduced load on the server. You
can move behavior in both directions: imagine, in the design in the
overhead, a server handling asynchronous computation-intensive
requests from some of the client applications.

Remote Method Invocation (RMI) 3-9

What Is Serialization?
Serialization allows you to save and restore the state of an object.
Without serialization, parameters and return values would have to be
simple primitive data types. The RMI architecture would not work as
seamlessly as it does without mechanisms for sending objects over the
network. This would limit drastically the usefulness of a distributed
design, and make the power of remote objects applicable only to trivial
applications.

Almost every application requires some way of keeping data. Most

applications use a database for the storage or persistence of data.
However, databases are not typically used to store objects, particularly
Java objects. A process is required to keep the state of a Java object in
such a way that the object can be easily stored and retrieved, and
returned to its original state.

Java object serialization is a lightweight object persistence technique

that allows you to convert objects into streams. The Java object
serialization framework consists of a set of methods that allow you to
duplicate an object by converting its field values into a stream that can
be easily reconverted to its original object representation by the
receiver of the stream.

3-10 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
The deserialized object is a copy of the original object and not the
original itself. To send an object reference to another object and not a
copy of the original, that is, pass the object by reference rather than by
value, use a Remote object, which means that the respective class
must implement the Remote interface.

Objects that act as containers to be serialized implement the interface

Serializable that allows the contents of the object to be saved or
restored in a single stream.

Remote Method Invocation (RMI) 3-11

Object Serialization Architecture

You can serialize almost all classes in the Java 2 SDK. The exceptions
are classes containing only static or transient fields, or both fields and
classes that represent specific characteristics of the local virtual
machine. There is no reason for these classes to be sent to other objects.

Classes that are made up of fields other than static or transient fields,
you can serialize, but the default mechanism does not make values of
the static or transient fields persistent. When you serialize the object,
the default values for the transient and static fields are set as if the
object would have been instantiated from scratch. For most RMI
applications, the default built-in serialization is all you need.

Object streams do not transmit an object’s bytecode; that is, the

compiled class files, instead, they transmit a representation of its
structure and related data.

3-12 Distributed Programming With Java Technology

Writing and Reading to and from an Object Stream

Writing to an Object Stream

Writing and reading an object to and from a file stream is a simple

process. For example, the following is a simple Point class:
1 class Point implements java.io.Serializable {
2 int x;
3 int y;
4 Point( int x, int y){
5 this.x = x;
6 this.y = y;
7 }
8 }

Remote Method Invocation (RMI) 3-13

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
The class Point implements the Serializable interface and can
therefore be written to an object stream with code that looks like the
following:
1 Point myPoint = new Point(1,2);
2 FileOutputStream fos = new FileOutputStream(myfile);
3 ObjectOutputStream oos = new ObjectOutputStream(fos);
4 oos.writeObject(myPoint);
5 oos.close();

You must add some exception handling, and the upper code in a try-
catch clause must be included as usual when dealing with
input/output (I/O) methods. The constructor for
ObjectOutputStream expects an OutputStream object.
FileOutputStream is a valid subclass allowing you to write the object
into a file stream.

Reading From an Object Stream

Reading the object is as simple as writing it, with one caveat—the

readObject method returns the stream as an Object type, and it must
be cast to the appropriate class before methods on that class can be
executed.
1 Point serialPoint;
2 FileInputStream fis = new FileInputStream(myFile);
3 ObjectInputStream ois = new ObjectInputStream(fis);
4 serialPoint = (Point)ois.readObject();

Time-Consuming Process

Serialization can be a time-consuming process. When an object is

serialized, the system is building up a graph representing the object
with all its contained references that can be traversed at the time of
deserialization. Complex objects result in complex graphs that take
time to construct.

3-14 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
Customizing readObject and writeObject

For special cases, the default serialization might not fulfill all needs.
Then you must customize the serialization according to your
requirements. You should be aware of the following rules:

● You must implement the interface java.io.Serializable.

● You must define a default constructor with no arguments.

● You must implement the readObject and the writeObject

methods.

Customizing the default methods could, for example, consist of

calculating values for special fields of the object. You can customize
the serialization only when the object is deserialized.

Transient and Static Fields

You cannot serialize fields that are marked transient or static. This
offers an easy way to prevent fields from being serialized. Changing a
non-static field to static also changes the semantics of this class. The
preferred way to protect a field from serialization without affecting its
behavior is to mark it as transient. Providing the methods
writeObject or readObject offers a way to set non-default values for
transient fields after deserializing the respective object.

Versioning Serializable Classes

Serializable classes define a version ID that helps to find out which

version of a class has been serialized and which version is expected for
deserialization. You can use the serialver utility to determine the
version ID. This utility comes with the Java 2 SDK, and returns the
serialVersionUID for one or more classes.

Remote Method Invocation (RMI) 3-15

Object Streams
Java object serialization produces just one stream format for the
encoding and storing of objects. Each object acting as a container
implements one of two interfaces: ObjectOutput or ObjectInput.
They define some methods for the reading or writing of objects and
primitives and some basic stream operations, such as close or flush.

All the work to be done with the serialization or deserialization of

objects is taken care of by the stream objects that implement either the
ObjectOutput or the ObjectInput interface.

ObjectOutputStream Class

The ObjectOutputStream implements the ObjectOutput interface,

which itself extends the DataOutput interface to write primitives. The
ObjectOutput provides an interface for object storage. The
writeObject method is used to write objects to a stream. An object is
checked first to see if it implements the Serializable interface before
it is written to the stream using the writeObject method. Exceptions
can occur while accessing the object or its fields, or when attempting to
write to the storage stream.

3-16 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
ObjectInputStream Class

The ObjectInputStream class is used for accessing objects written to

an ObjectOutputStream, which implements the interface
ObjectInput. This in turn extends DataInput to read primitives. The
readObject method defined in the interface is used for object
retrieval. It reads from the storage stream and returns an object. The
readObject first gets an identification header from the stream. The
next step is to read the bytecode and to create an instance with the
default constructor before the readObject method fills in the object’s
value fields. Exceptions are thrown when attempting to read the
storage stream, or if the class name of the serialized object cannot be
found.

Serializable Interface

Any object that is serializable must implement the Serializable

interface or implement a class that extends the Serializable
interface. This interface is in fact empty and serves merely to flag
objects that use serialization.

All of the fields (data) of a Serializable object are written to the

storage stream. This includes primitive types, arrays, and references to
other objects. Again, only the data (and class name) of the referenced
objects are stored. Methods are not written, because methods are
assumed to be unchanged and can be accessed from the restored object
state.

Remote Method Invocation (RMI) 3-17

The Externalizable Interface

For classes that implement the Externalizable interface, the
container saves only the identity of the class. In this case, the storage
and retrieval of the object’s state become the responsibility of the
object itself.

Externalizable objects must:

● Implement the java.io.Externalizable interface.

● Implement a writeExternal method to save the state of the

object. The method must explicitly coordinate with the supertype
to save its state.

● Implement a readExternal method to read the data on the stream

and restore the state of the object. The method must explicitly
coordinate with the supertype.

● Be responsible for the externally defined format. The

writeExternal and readExternal methods are solely
responsible for this format.

3-18 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
Exercise: Serialization

Exercise objective – Serialize and deserialize a class, and test the

versioning mechanism of serialization.

Tasks

Complete Templates to Deserialize and Reserialize a Given Class

In this exercise you complete two templates: Serialize.java and

Deserialize.java. These classes are used to serialize and deserialize
the class Name.java, which already exists.

1. Change the directory to labfiles/mod3-rmi/lab/serial.

2. Modify the template Serialize.java to serialize the class

Name, and write the output into a file stream. The class Name
already exists in the directory. Compile your class (pay close
attention to the package structure).

3. Start Serialize with the following command:

java Serial.Serialize test.ser Firstname Lastname

By doing this, you provide three String parameters: the name of

the file with the serialized object (test.ser), a first name, and a
second name.

4. Modify the template Deserialize.java so that it reads the

serialized Name class from the file test.ser and displays the
two strings on the console.
✓ The deserializing command should be java Serial.Deserialize test.ser.

Remote Method Invocation (RMI) 3-19

Find the Version of a Class and Test the Versioning Mechanism

Find the version of Name.class, by using the serialver utility

provided with the Java 2 SDK.

1. Show the version number of the class Name.class with the

serialver utility.
✓ Remember the package. The correct command for step 5 is serialver Serial.Name.

2. (Optional) Modify Name.java in a way that changes its

serialver. Then try to deserialize an “old” test.ser file.

3. (Optional) Modify your “new” Name.java by setting its

serialver manually to the serialver of the “old” Name.java.
(Use the output of the serialver utility from step 5.) Then try
again to deserialize an “old” test.ser file.

3-20 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
Exercise Summary
Discussion – Take a few minutes to discuss what experiences, issues,
or discoveries you had during the lab exercises.

● Experiences
✓ Ask students what their overall experiences with this exercise have been. You may want to
go over any trouble spots or especially confusing areas at this time.

● Interpretations
✓ Ask students to interpret what they observed during any aspects of this exercise.

● Conclusions
✓ Have students articulate any conclusions they reached as a result of this exercise
experience.

● Applications
✓ Explore with students how they might apply what they learned in this exercise to
situations at their workplace.

Remote Method Invocation (RMI) 3-21

Creating an RMI Application

The above overhead shows graphically the necessary steps to create an

RMI application. The red (or dark) parts represent code that you must
write, and the green (or light) parts represent code that is
automatically generated with the RMI compiler utility.

3-22 Distributed Programming With Java Technology

Steps to Create an RMI Application

There are six major steps involved in creating the RMI application:

1. Develop the Java technology interface (“Java interface”) that

defines the remote object.

2. Create the implementation class (the servant) for the interface.

3. Create the server, which manages the servant instances.

4. Create the client, which uses the remote object (in the end, the
client uses the servant).

5. Use the Java compiler to create the Java class files.

6. Run the RMI compiler (rmic) to create the helper classes (stubs
and skeletons).

Remote Method Invocation (RMI) 3-23

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
Step 1: Develop the Java Interface

The Java interface describes the server object. This interface has two
differences, as compared to a “regular” Java interface: the Echo
interface must extend java.rmi.Remote, and every method must
throw java.rmi.RemoteException.

The following is an example of the interface describing an Echo object:

1 package EchoApp;
2 import java.rmi.Remote;
3 import java.rmi.RemoteException;
4
5 public interface Echo extends Remote {
6 public String sayEcho(String myName)
throws RemoteException;
7 }

Step 2: Create the Implementation Class (Servant)

For each interface, you must write a servant class. Instances of the
servant class implement remote objects. Each instance implements a
single remote object, and each remote object is implemented by a
single servant. The servant must extend the class
java.rmi.server.UnicastRemoteObject

The servant provides an implementation for all of the methods

declared in the interface. In this case it is just one: sayEcho. In
addition, if the class does not have a constructor, you must explicitly
provide a modified default constructor. If you do not, the compiler
does not “like” the class and issues an error message.

3-24 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
Despite the fact that the implementation classes are called servants, the
usual naming convention is to append Impl to the Java interface name:
EchoImpl.java.
1 package EchoApp;
2
3 import java.rmi.RemoteException;
4 import java.rmi.server.UnicastRemoteObject;
5
6 public class EchoImpl
extends UnicastRemoteObject
implements Echo {
7
8 public EchoImpl() throws RemoteException {};
9
10 public String sayEcho(String myName) {
11 return "\nHello " + myName + "!!\n";
12 }
13 }
✓ This sayEcho method does not throw RemoteException, even if the declaration in the
interface Echo.java says so. Why? First, the compiler chooses not to throw an exception,
since doing this does not break the contract between the interface implementation class
and the interface user class. Second, the servant never throws a RemoteException on its
own, so why should there be a declaration which suggests that it will? The
RemoteException is, in fact, generated and thrown by the client side stub if something
goes wrong in one of the communication layers.

Step 3: Create the Server

The RMI runtime uses the server to instantiate and publish the
servant. In its simplest form, the server pre-instantiates all the required
servants and keeps them alive forever (or at least as long as the server
lives). In a production system this would usually not be good enough.
To keep thousands of servants alive, even when they are not actively
used by clients, is a waste of resources. In this case, more complex
servers must be developed and used. They must be able to instantiate
servants on demand, and maybe even persist and discard them after a
period of non-usage. The server, EchoServer.java, is a simple server:
1 package EchoApp;
2
3 import java.rmi.Naming;
4
5 public class EchoServer {
6 public static void main(String args[])
throws Exception{

Remote Method Invocation (RMI) 3-25

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
7
8 // Create the servant
9 // instance for registration
10 EchoImpl echoRef = new EchoImpl();
11
12 // Bind the object to the rmiregistry
13 Naming.rebind("Echo", echoRef);
14
15 System.out.println
("Echo object ready and bound
to the name 'Echo'!");
16 }
17 }

You have to instantiate the servant and register it with the name server
(which is part of rmiregistry). After the servant is registered, its
object reference can be retrieved and used by the client application.
The following section describes how this works.

3-26 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
Step 4: Create the Client Application

The following describes how to create the client application.

1 package EchoApp;
2
3 import java.rmi.Naming;
4
5 public class EchoClient {
6 public static void main(String args[])
throws Exception {
7
8 //Check the argument count
9 if (args.length != 2) {
10 System.err.println("Usage:
EchoClient <server> <your name>");
11 } else {
12 //Get the Hello instance,
13 //cast it to the Hello interface
14 String url = new
String("rmi://"+args[0]+"/Echo");
15 Echo echoRef = (Echo)Naming.lookup(url);
16
17 // call the Echo server object
18 // and print results
19 String reply = echoRef.sayEcho(args[1]);
20 System.out.println(reply);
21 }
22 }
23 }

Echo was used as a name to bind the servant into the name server.
Now use the same name to look it up. The result is an object of type
java.lang.Object, which you have to cast to an Echo to be able to
use its methods.

Step 5: Compile the Java Class Files

Use the following command to compile the files:

javac -d . *.java
✓ Run this command from the /mod3-rmi/solutions/echo/ directory.

Remote Method Invocation (RMI) 3-27

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
Step 6: Create the Stubs and Skeletons

To complete the setup, you need the stub and the skeleton class.
Provide a stub and a skeleton for every class that implements the
java.rmi.Remote interface. In this case, this is EchoImpl.java,
which implements Echo.java, which extends Remote. The stub
compiler runs against the previously compiled class file.

The utility to create the stub and skeleton is called the RMI compiler or
rmic. The syntax is similar to the javac compiler. The stub and
skeleton are compiled by typing the following:
rmic -d . EchoApp.EchoImpl

In your echo directory, you should now have an additional EchoApp

directory, which contains the following files:
Echo.class
EchoClient.class
EchoImpl.class
EchoImpl_Skel.class
EchoImpl_Stub.class
EchoServer.class

If this is not the case on your system, redo steps 1–6 or ask your
instructor what could have gone wrong.

3-28 Distributed Programming With Java Technology

Deploying an RMI Application

First Try: Start Everything Local

Assuming there are no compilation problems, the RMI registry is

started first, then the server, and finally the client.
rmiregistry (Windows: start rmiregistry)

The RMI registry does not produce any output, so you can safely start
it in the background.

The server, which creates the servant and registers it with the RMI
registry, is started next.
java EchoApp.EchoServer

The server does produce some output, so it is better to start it in its

own console or command prompt.

In the last step, start the client.

java EchoApp.EchoClient localhost <your name>

Remote Method Invocation (RMI) 3-29

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
Second Try: Connect to a Remote System

To connect your client to a different machine, provide the name of the

machine running the RMI registry you want to use.
java EchoApp.EchoClient <host name> <your name>

This line connects you to host <host name>. Because a server can
bind servants only to an RMI registry running on the same host as the
server itself, <host name> also runs the servant.

Right now, all the files necessary to start up the server, servant, and
client are accessible in the current directory or through the class path.
You will see in a later section how to download the necessary code on
demand.

3-30 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
Exercise: Compiling a Basic RMI Application

Exercise objective – Compile and run the Echo application. In

addition, test the robustness of the RMI system by breaking the code
of the Echo application intentionally.

Tasks

Compile the Echo Application

Work through the six steps described in the handout to compile the
Echo application.

1. Change the directory to labfiles/mod3-rmi/lab/echo.

2. Compile the entire Echo application using javac.

3. Compile the stub using rmic.

✓ You can find more detailed instructions in the README file included in
/mod3-rmi/solutions/echo/.

Run the Echo Application

Work through the steps described in “Deploying an RMI Application”

to get the Echo application running.

1. Run the Echo application locally on your system.

2. Team up with somebody, and distribute the client and the server
parts of Echo on two different machines.

Break the Echo Application to Test the Robustness of the RMI

System

After Echo runs successfully, try to break the code in several ways to
test the fault tolerance of RMI. Try out the following in the servant
code (EchoImpl.java):

Remote Method Invocation (RMI) 3-31

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
3. Do not implement a method that is in the interface (or fail to
implement the correct method signature).
✓ The code does not compile successfully.

4. Omit extends UnicastRemoteObject.

✓ The code can be compiled and rmic runs without error. However, at runtime a
MarshalException gets thrown.

5. Omit implements Echo.

✓ The code can be compiled successfully, but rmic complains. Class EchoApp.EchoImpl
does not directly implement a Remote interface. This is true, since Echo.java extends
Remote, but Echo.java was no longer implemented.

Work With the RMI Online Documentation (Optional)

Use the online RMI documentation to solve a specific problem.

1. Sometimes, you have your own class hierarchy, and cannot

extend UnicastRemoteObject with your servant
implementation. Look at the documentation for
java.rmi.server.UnicastRemoteObject to find out how you
can subclass from an arbitrary object, and still create a valid
remote object.

2. Change HelloImpl.java to subclass Object instead of

UnicastRemoteObject.

3-32 Distributed Programming With Java Technology

● Experiences
✓ Ask students what their overall experiences with this exercise have been. You may want to
go over any trouble spots or especially confusing areas at this time.

● Interpretations
✓ Ask students to interpret what they observed during any aspects of this exercise.

● Conclusions
✓ Have students articulate any conclusions they reached as a result of this exercise
experience.

● Applications
✓ Explore with students how they might apply what they learned in this exercise to
situations at their workplace.

Remote Method Invocation (RMI) 3-33

RMI Architecture

RMI Architecture Overview

An RMI application consists of four layers:

● Application

● Stub/skeleton

● Remote reference

● Transport

The client code first talks to its stub, which then sends the message to
the remote reference layer (RRL). The RRL then passes it through the
transport layer to the server machine.

At the server, the message goes all the way up again. The transport
layer passes it to the RRL, which in turn retranslates it to the skeleton
where it finally appears at the server’s object implementation.

3-34 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
As a developer, you are responsible for creating only the Java RMI
interface definitions and the implementation classes and for the
generation of the stub and skeleton classes. The RRL and transport
layer implementations are taken care of for you.
✓ Skeletons are not required by Java 2. They are required for the Java 1.1 RMI model.
Skeletons are required if compatibility between Java 2 and Java 1.1 is desired (Java 2, The
Complete Reference, Patrick Naughton and Herbert Schildt, McGraw-Hill, 1999, p. 836).

Remote Method Invocation (RMI) 3-35

Invocation Overview
Local Java objects are invoked by reference; however, the RMI system
is passing objects in parameters or return values by copy and not by
reference, because a reference to a local object makes sense only within
one virtual machine. RMI uses the Java programming language
serialization mechanism to flatten a local object into a serial stream
and send it over the wire to the receiving instance.

The remote methods are described by interfaces that are in turn

implemented by remote objects called servant objects. At least one
interface that extends the java.rmi.Remote interface must be
implemented by the remote object. The remote object is accessible only
to other objects through the methods defined in its interface.

3-36 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
RMI clients interact with remote objects using their published
interfaces. The actual object is not sent to the client. The client gets a
proxy or stub; that is, a handle to the remote object, by looking up the
registry or by the return value of another remote method call. It is this
stub the client is interacting with when remote methods are invoked.
A single remote object can have many clients, which in turn all have
their own stubs representing the remote object. On the server, a similar
structure exists. A server-side proxy called a skeleton is responsible for
passing the method calls to the referenced object.

The Java 2 SDK can get rid of the skeleton class. This does not imply
that a skeleton is no longer needed. The RMIRemoteObject class has
been modified and now has the additional job of being a universal
skeleton, in combination with the Java programming language
introspection mechanism.

The invocation of a method on a remote object is definitely not the

same as calling a local object, but the code looks similar and can easily
be understood. Nevertheless, keep the following differences in mind:
● Objects that are passed as parameters to remote methods and
objects that are returned from a method are passed by value rather
than by reference. A reference to the remote object is used only if a
remote object is passed to or returned by a remote object.
● Objects that are passed as parameters to remote methods, and
objects that are returned from a method, must be serializable.
● Remote objects are referred to by clients through the implemented
remote interface. Casting the remote object to any of the
implemented interfaces is therefore possible.
● The toString method has been changed to include information
about the used network protocol and the originating host name
and port number.
● The hashCode method has been modified to return identical keys
for any references to the same remote object.
● The equals method checks whether the object references of
remote objects are equal and not the contents of remote objects.

Remote Method Invocation (RMI) 3-37

Interface Descriptions
You can use RMI between virtual machines on a single system or over
a network between virtual machines (VMs) on multiple machines.
Using interfaces allow you to create a highly modular application
design.

An application component is designed as a set of well defined

interfaces that objects use to communicate with each other. That is, the
interaction with an object is entirely described by the interface. The
actual design and implementation of these components is invisible
outside of the individual object and can be replaced anytime without
any restrictions for the rest of the system.

In RMI, all interfaces are described in pure Java code. With the RMI
stub compiler (rmic) the stub and skeleton files are generated directly
from the compiled code that implements the interfaces.

3-38 Distributed Programming With Java Technology

The Application Layer

The application layer consists of the actual implementation of the
client and server applications. The methods that are made available for
remote clients are described in an interface that extends
java.rmi.Remote. The java.rmi.Remote interface does not contain
any methods and is used only to mark objects as remotely accessible.
Such interfaces are implemented as usual; only some additional code
for dealing with RemoteExceptions is needed.

As a subclass of the class UnicastRemoteObject, the export of the

remote object is handled by the superclass. You can also call the
exportObject method explicitly.

The final part for an RMI application is to register the remote object
with a name server to be available for the first contact of a requesting
client.

Remote Method Invocation (RMI) 3-39

The Stub and Skeleton Layer

The stub as the client-side and the skeleton as the server-side
representation are class files that are generated by the RMI stub
compiler (rmic). This layer does not deal with any of the specifics of
any transport, but transmits data to the remote reference layer.

The stub is responsible for the initiation of calls to the remote object it
represents. It is also the stub that marshals method arguments to a
stream object that contains parameters, errors, and exceptions, and
unmarshals (receives and interprets) streams returned by the remote
object.

All interfaces of a remote object are also implemented by the stub class
that represents this object at the client. A stub class is type equivalent
to any of the represented remote interfaces and can therefore be cast as
any of those. A test with the instanceof operator answers whether a
certain interface is implemented by a remote object.

3-40 Distributed Programming With Java Technology

Stub Communication
Stubs interact with the client-side RRL in the following ways:

● The stub receives the remote method invocation and initiates a call
to the remote object.

● The RRL returns a marshal stream, which is used to communicate

with the server’s RRL.

● The stub makes the remote method call, passing any arguments to
the stream.

● The RRL passes the method’s return value to the stub.

● The stub acknowledges to the RRL that the method call is

complete.

Remote Method Invocation (RMI) 3-41

Skeleton Communication
The skeleton class receives the method calls, marshals the parameters,
and dispatches the methods to be exported. Skeletons interact with the
server-side RRL in the following ways:

● The skeleton unmarshals (receives and interprets) any arguments

from the I/O stream, established by the RRL.

● The skeleton makes the up-call to the actual remote object

implementation.

● The skeleton marshals (interprets and sends) the return value of

the call (or an exception, if one occurred) onto the I/O stream.

As mentioned before, the skeleton class is no longer needed with the

Java 2 SDK, its function is integrated in UnicastRemoteObject.

3-42 Distributed Programming With Java Technology

The Remote Reference Layer

The remote reference layer (RRL) bridges the gap between stub and
skeleton code and the network communication.

The RRL is responsible for carrying out the semantics of the method
invocation. It manages communication between the stubs/skeletons
and the lower-level transport interface using a specific remote
reference protocol, which is independent of the client stubs and server
skeletons. The RRL’s responsibilities include managing references to
remote objects and reconnection strategies if an object should become
unavailable.

The RRL has two cooperating components: the client side and the
server side. The client-side component contains information specific to
the remote server, and communicates using the transport layer to the
server-side component. The server-side component implements the
specific remote reference semantics prior to delivering a remote
method invocation to the skeleton.

Remote Method Invocation (RMI) 3-43

The Transport Layer

The transport layer is responsible for connection setup, connection
management, and keeping track of and dispatching remote objects (the
targets of remote calls) residing in the transport’s address space.

The transport layer performs the following tasks:

● Receives a request from the client-side remote reference layer

● Locates the RMI server for the remote object requested

● Establishes a socket connection to the server

✓ Currently, Java RMI uses TCP sockets. Typically, there are two socket connections
between the server and client: one for method calls and one for the DGC. RMI will attempt
to re-use existing socket connections for multiple objects from the same server, but if an
existing socket is in use when the attempt is made, a new socket will be created.

3-44 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
● Passes that connection back up the client-side remote reference
layer

● Adds this remote object to a table of remote objects with which it

knows how to communicate

● Monitors connection “liveness”

✓ A client cannot close a connection to an RMI server because it is handled at this layer, so
1.1 connections time out if they are unused for a period of time. This period defaults to 10
minutes, and is set by the java.rmi.dgc.leaseValue property.

You can adjust the transport layer to deal with special requirements,
such as encryption or data compression.

Remote Method Invocation (RMI) 3-45

Garbage Collection
RMI uses a reference-counting garbage collection scheme that keeps
track of external live references to remote objects within a local virtual
machine. A live reference is, in this case, an active connection over a
TCP/IP session.

Distributed garbage collection is a tricky field. For example, a remote

object on the server is prematurely collected as garbage because of a
link failure. The client tries to reconnect with a reference to an object
that does not exist anymore. This causes a RemoteException that
must be handled by the calling object. There is no automatic rebinding.

When a live reference enters a JVM, its reference count is incremented.

When an object is found to be unreferenced, the reference count is
decremented as the object is finalized. When the last reference has
been discarded, an “unreferenced” message is sent to the server.

A remote object that is no longer referenced by any client is considered

to have a weak reference. The weak reference allows the server’s
garbage collector to discard the object if no other local references to the
object exist. As long as a local reference to a remote object exists, it
cannot be collected as garbage.

3-46 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
A reference to a remote object is in fact leased by the client and ends
after a certain time. This period defaults to 10 minutes and can be set
by the java.rmi.dgc.leaseValue property. The client starts
leasing with a call and is responsible for extending the leasing
periodically by additional calls. If the client fails to make the
additional calls, the leasing of the remote reference expires and the
garbage collector will do its job. For local references, the garbage
collector calls the finalize method of the object before removing it
from memory, which holds a chance for a last notification.

Remote Method Invocation (RMI) 3-47

Garbage Collection
The steps involved in garbage collection are:

1. The server (implementation) starts and creates an object that is

referenced remotely. This establishes a weak reference to the
object.

2. When the client requests the object, the client’s JVM creates a live
reference, and the first reference to the object sends a
“referenced” message to the server.

3. When the object goes out of scope on the client, an

“unreferenced” message is sent to the server.

4. When the count on the object goes to 0, and there are no local
references to the object, the object reference can be passed to the
server’s local garbage collector for collection.

3-48 Distributed Programming With Java Technology

RMI Object Hierarchy

The Remote interface identifies remote objects. Every remote object
must implement this interface, which itself does not contain any
methods. Its purpose is only to flag remote objects.

The RemoteObject class acts like a remote representative for the Java
root class Object and provides some overridden methods in their
distributed version, such as hashCode, equals, and toString. The
getClass method works for local as well as for remote objects. Used
with a remote object, it returns the stub’s type.

The RemoteServer class is used for the creation and export of server
objects and is therefore the superclass of all RMI servers. The
getClientHost method returns the IP address of the requesting client.
The getLog and setLog methods allow logging.

Most servants are subclassing UnicastRemoteObject. References to a

server object are valid only during the lifetime of this object. The
exportObject method allows a stub object to be exported explicitly.

Remote Method Invocation (RMI) 3-49

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
The RemoteStub class is the superclass for all client stubs. It is the
client’s representative of an implementation class.

The RMI system currently does not support multicast remote objects
and replication.

3-50 Distributed Programming With Java Technology

RMI Naming Service

The RMI registry enables remote objects to be retrieved and registered

using simple names. Each server process can either maintain its own
registry or be part of a single registry that supports all virtual
machines on a local system.

Remote objects that are to be exported have to register their names to

be found by a requesting client. The server application uses the static
bind, unbind, and rebind methods of the java.rmi.Naming class to
register its object implementations with the RMI registry. A URL-based
lookup by the client that also uses the Naming class to achieve this can
then be performed when the client requests a reference to a remote
object.

The normal registration process is usually handled by maintaining the

RMI registry on a defined port number allowing a client to query the
service with the lookup or list methods. An alternative design
would be to integrate the naming services into the RMI application to
gain full control over the registry.

Remote Method Invocation (RMI) 3-51

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
The fact that remote objects can return other remote objects means that
the initial contact to the registry service—the bootstrapping of the
naming service—is performed only once. Further references to remote
objects on the respective server can be obtained by calling the already
referenced object.

In the Echo application, the naming service is reflected in the RMI

client class (EchoClient.java) with its lookup method and in the
implementation class
// EchoClient.java:
...
String url = new String("rmi://"+args[0]+"/Echo");
Echo EchoRef = (Echo)Naming.lookup(url);
...

and in the RMI server class with the binding to the Naming class.
// ;
// EchoServer.java:
...
// Create the servant instance for registration
EchoImpl echoRef = new EchoImpl();

// Bind the object to the rmiregistry

Naming.rebind("Echo", echoRef);

The rebind method associates or “binds” the name Echo to the

echoRef object, removing any object previously bound with this name
from the registry.

The Naming class provides two static methods that enable the
developer to register an implementation, bind and rebind. The only
difference is that the bind method throws a
java.rmi.AlreadyBoundException if another object has already
been registered on this server, using the name passed as the first
argument to the method.

3-52 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
The arguments to bind and rebind are a URL-like String and the
instance name of the object implementation. The String is expected
to be in the format:

rmi://host:port/name

where rmi is the protocol, host is the name of the RMI server (which
might need to be fully DNS or NIS+ qualified), port is the port
number that the server should be listening to for requests, and name is
the exact name that clients should use in Naming.lookup requests for
this object. If not specified, the protocol defaults to rmi, host defaults
to the local host, and port defaults to 1099.
✓ For security reasons, an application can bind only to a registry running on the same host.

The rmiregistry Application

The rmiregistry is an application that provides a simple non-
persistent naming lookup service that helps a client to locate remote
objects. The EchoServer provides rmiregistry with the object
reference and a String name through the rebind method call.
✓ The registry is also responsible for polling the table of remote objects, and providing
reference information (if each object has a live reference or a weak reference) to the
distributed garbage collector.

The rmiregistry must be running before the EchoServer application

attempts to bind.
rmiregistry

You can set properties for the RMI server JVM from the command line:

● java.rmi.server.codebase – A URL that indicates from where

clients can download classes.

● java.rmi.server.logCalls – If set to true, the server logs calls

to stderr. By default it is set to false.
java -Djava.rmi.server.logCalls=true EchoApp.EchoServer
&

Remote Method Invocation (RMI) 3-53

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
Once the implementation is exported by the registry, the client can
send a URL string to request that the rmiregistry provide a
reference to the remote object. The lookup is accomplished through a
client call to Naming.lookup, passing in a URL string as the
argument:
rmi://host:port/name

3-54 Distributed Programming With Java Technology

Catching Exceptions
There are many more ways for distributed applications to fail than
there are for their local counterparts. Consequently, the RMI systems
extend the Java exception classes to deal with the additional
complexity. The exception handling is therefore not bound to a local
virtual machine.

There are various reasons for exceptions to be thrown during a remote

method invocation. Consequently, there are various exceptions defined
in the RMI system to reflect this. The RemoteException class is the
superclass of all RMI exceptions.

In the Echo application, there are try-catch clauses in the client class
embracing the lookup method, as well as in the server class around
the code that binds the object to the registry.

Remote Method Invocation (RMI) 3-55

Remotely Loaded Code

The Java programming language allows you to download code from a

remote system, such as from a Web server. This is also true for applets
doing RMI. Keep in mind that an RMI applet needs more than just the
applet class, there is also the stub and the interface. These classes also
must be available on the HTTP server.

With RMI, you can download some of the required classes for an RMI
application from a Web server. There is a special class called
RMIClassLoader, which is part of the rmi.server package. This class
loader is used by the RMI runtime system transparently if it cannot
find a required class in the local class path. The most important class
that does not have to be on the client is the stub. Other new classes
could be acquired as a return value from a remote call. If all the client
“knows” about a return value is the interface, the actual class that
implements this interface is known only at runtime.

3-56 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
The java.rmi.server.codebase Property
While the server gets all its needed classes from the class path, the
client has to get them from a Web server. How does the client know
which URL to use? You could hardcode the URL in the client code, or
deliver it as a startup argument on the client side. The problem with
these approaches is that they make it difficult to relocate a server. All
clients or client startup scripts would have to be updated.

RMI offers a better approach. The URL is set on the server side, and
RMI clients learn about the URL automatically; it is embedded in the
RMI wire protocol. The URL is set as a system property when the
server is started. For example:

java -Djava.rmi.server.codebase=http://myhost:8080/
EchoApp.EchoServer

This line tells the clients to get their class files from the host galahad on
port 81. The trailing “/” is mandatory, because the URL always
denotes a directory, and never a single file.

Note – The rmiregistry application also gets the URL from the
server. At the moment the server binds the servant into the registry,
the rmiregistry needs the class file. If the rmiregistry is able, at
this moment, to locate the class using the class path, instead of using
the server supplied URL, the URL system ceases to work. A client
cannot get the class using the URL anymore. It is unclear why this is
the case, but this behavior causes a lot of confusion about the
java.rmi.server.codebase property. If you start the rmiregistry,
be sure to start it with an empty class path.

Remote Method Invocation (RMI) 3-57

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
Security Aspects

Security on the Client

It is potentially dangerous to download remote code. Therefore, to use

the RMIClassLoader, you must have a SecurityManager in place
to ensure that classes loaded from the network meet the desired
security restrictions. If there is no security manager running, the
application can load classes only from the class path. You do not have
to develop a SecurityManager on your own, RMI provides one: the
RMISecurityManager. This security manager behaves much like the
AppletSecurityManager that is present in most browsers.

Security on the Server

The typical closed-system scenario has the server configured to load

no remote classes. The services it provides are defined by remote
interfaces that are all local to the server machine. The server has no
security manager and cannot load classes even if clients send along the
URL. If clients send remote objects for which the server does not have
stub classes, those method invocations fail when the request is
unmarshalled, and the client receives an exception.

The more open server system defines its

java.rmi.server.codebase so that classes for the remote objects
it exports can be loaded by clients, and so that the server can load
classes when needed for remote objects supplied by clients. The server
has both a security manager and RMI class loader, which protects the
server. A somewhat more cautious server can use the property
java.rmi.server.useCodebaseOnly to disable the loading of
classes from client-supplied URLs.

A client can supply a URL in the same way a server does, by setting
the java.rmi.server.codebase property on the client.

3-58 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
Exercise: Remotely Loaded Code

Exercise objective – So far, all the code needed to run the Echo client
and server was loaded from the current directory or from the class
path. Change this so that the client receives its stub automatically
using a Web server.

Preparation
With the lab setup comes a mini Web server that is precompiled. Start
the mini Web server, and pass the port number where it should serve
the files, and the path where the EchoApp package is found after you
complete it.

java httpsrv.ClassFileServer 8080 ~/labfiles/echoremote

✓ Compile from ~/SL301_revC_August_1999/SL301_XXX_LF using the following command
javac httpsrv/*.java , then launch the server java httpsrv.ClassFileServer 8080
~/SL301_revC_August_1999/SL301_XXX_LF/labfiles/echoremote

Tasks

Modify the Client so That it Can Receive Code From a Web Server

The Java programming language allows you to load code only from a
remote location if a security manager is in place. The first task is to
modify the client to instantiate and install a security manager.

1. Change the directory to labfiles/mod3-rmi/lab/echoremote.

2. Modify the client to instantiate and install an

RMISecurityManager. Use the Java 2 SDK documentation to find
out how to accomplish this (look specifically at the classes
java.rmi.RMISecurityManager and at java.lang.System).

Remote Method Invocation (RMI) 3-59

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
Move the Compiled Stub out of the Client’s Class Path

The class path is always searched before any remote location is tried.
Therefore, you must move the client code so that the stub is no longer
in the class path.

1. In the echoremote directory, you will find an empty directory,

client. Use this directory as a starting point to launch the client.
Copy the client (EchoClient.java) and the interface
(Echo.java) into this directory. (Be sure to preserve the package
structure.)

2. Start the rmiregistry in such a way so that none of the Echo

classes are in the class path.

3. Launch the server and pass the location of the Web server in the
java.rmi.server.codebase property.
java -Djava.rmi.server.codebase=http://yourhost:8080/
EchoApp.EchoServer

There is a blank just before EchoApp.EchoServer.

4. Now you are ready to test the client. Type:

java EchoApp.EchoClient yourhost Emerald

The client should still work, even if the stub is no longer available
from the class path.

Test the Robustness of the Remote Code Loading Mechanism

(Optional)

Once you are convinced the download of code works, try to break it to
find out how robust the system is. Try the following:

1. Do not load the RMISecurityManager.

2. Start the rmiregistry.

3. Do not set the java.rmi.server.codebase property.

4. Set the property on the client instead of on the server.

3-60 Distributed Programming With Java Technology

● Experiences
✓ Ask students what their overall experiences with this exercise have been. You may want to
go over any trouble spots or especially confusing areas at this time.

● Interpretations
✓ Ask students to interpret what they observed during any aspects of this exercise.

● Conclusions
✓ Have students articulate any conclusions they reached as a result of this exercise
experience.

● Applications
✓ Explore with students how they might apply what they learned in this exercise to
situations at their workplace.

Remote Method Invocation (RMI) 3-61

Advanced RMI

Object Factories
A servant must be running to be used by its clients, but a client cannot
start a servant directly (this is the job of the server). The technique that
defines this limitation is called object factory and allows the creation of
multiple instances of remote objects on the fly, controlled by the client
application.

The basic idea of object factories is to have distributed constructors to

create the needed objects. An object factory is therefore a class that has
one goal: create instances of objects with which clients need to interact.
Normally, object factories only have a single “create a new object”
method that is called by a client. The calling object gets a remote
reference back to this newly created instance and can then call
methods on it.

The following sections describe how the Echo application is converted

to use an object factory.

3-62 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
Creating the Factory Interface

There is no need to change the original Echo interface, because the

sayEcho method of the original servant is still needed. Instead, you
must develop a new interface: the interface for the object factory. It
describes the methods a factory server must implement to create other
server objects. There must be at least one method doing two jobs:
create a new servant instance, and return an object reference to this
new instance.

In the Echo example, the factory looks like the following:

1 public interface EchoFactory extends Remote
2 Echo getEcho(String greeting) throws
3 RemoteException;
4 }

The getEcho method returns—as its name suggests—an instance of

Echo. The getEcho method also takes a greeting string which is used
to customize the different remote Echo instances.

Creating the Factory

The object factory is a simple RMI server class extending

UnicastRemoteObject and implementing the EchoFactory interface;
that is, it must provide the body for the getEcho method.
1 public Echo getEcho(String message)throws
2 RemoteException {
3 EchoImpl echoRef = new EchoImpl(message);
4 return (Echo)echoRef;
5 }

The cast from EchoImp to Echo in line 3, which is needed to fulfill the
return value defined in the interface definition. With this factory
implementation, a client can create different remote Echos
dynamically.

Remote Method Invocation (RMI) 3-63

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
Adjusting the Echo Implementation File

Theoretically, the EchoImpl could have been left unchanged, because

the Echo interface did not change. But the problem with the current
EchoImpl is that one EchoImpl instance is indistinguishable from
another. There is no point in having more than one instance if all the
instances do the same task. This is why EchoImpl is changed to
support the customization of its greeting message. This way, different
EchoImpl instances can be constructed, each one greeting in a different
language. To accomplish this, you add a new constructor. The new
constructor takes the greeting string and stores it internally. Not
shown are the changes to the sayEcho method, which replaces the
hardcoded message with the contents of the greeting variable.
1 public EchoImpl(String greeting) throws
2 RemoteException {
3 this.greeting = greeting;
4 }

Changing the Server

The Echo server does the same task. It creates a servant that is up and
running and waiting to be called. However, the servant is the factory.
The factory then is in charge of instantiating instances of Echo.
EchoFactory factoryRef = new EchoFactoryImpl();

The EchoFactory is bound to the registry as usual:

Naming.rebind("EchoFactory",factoryRef);

Changing the Client

The client now has the additional job of requesting one or more Echo
instances. But first, it needs a reference to the remote factory object
from the registry:
EchoFactory remoteFactory =
(EchoFactory)Naming.lookup(url);

3-64 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
It then applies the getEcho method several times using different
greeting strings and stores the references it gets back:
1 Echo echoRef1 = remoteFactory.getEcho("Hello");
2 Echo echoRef2 = remoteFactory.getEcho("Hi");
3 Echo echoRef3 = remoteFactory.getEcho("Howdy");

The sayEcho method is then called on these instances and the

resulting string is displayed on the client console:
1 String reply = echoRef2.sayEcho(args[1]);
2 System.out.println(reply);
3 reply = echoRef1.sayEcho(args[1]);
4 System.out.println(reply);
5 reply = echoRef3.sayEcho(args[1]);
6 System.out.println(reply);

Remote Method Invocation (RMI) 3-65

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
Exercise: Object Factory

Exercise objective – Implement the changes described in the Object

Factory section.

Tasks

Modify the Echo Application to Match the Setup Described in the

Object Factory Section

Complete the following steps:

1. Change the directory to labfiles/mod3-rmi/lab/factory.

2. The goal is to build an object factory that delivers Echo instances

doing the same thing as the original Echo servants: they
complete a string and send it back to the client. To achieve this,
some of the existing files you must modify, and write two
additional files: a factory interface (EchoFactory.java) and its
implementation (EchoFactoryImpl.java). Consider which of
the original files you must modify.

3. Create the factory interface (EchoFactory.java). It should

consist of single method getEcho(String greeting) that
returns a variable of type Echo and takes as parameter a String
variable greeting. Do not forget to throw the
RemoteException. Later, this method is called with different
greetings to get specialized remote Echo objects.

4. Create the implementation for the new factory interface. Take the
EchoImpl.java as a model. Name the new file
EchoFactoryImpl.java. The goal of the getEcho method is
to create new instances of objects that implement the original
Echo interface and to return a handle to this new instance.

5. Add another constructor to the EchoImpl.java file that accepts

the greeting string when an instance is created by the getEcho
method of the object factory.

3-66 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
6. Change the old EchoServer.java. It should now start the
factory, and no longer directly start an Echo.

7. Enable EchoClient.java to get the remote factory object from

the registry and then apply the getEcho method several times
using different strings, such as “Hello” and “Ciao” to have the
factory create some objects. As in the original EchoClient code,
the sayEcho method is called on these instances and the resulting
string is displayed on the console.

Remote Method Invocation (RMI) 3-67

● Experiences
✓ Ask students what their overall experiences with this exercise have been. You may want to
go over any trouble spots or especially confusing areas at this time.

● Interpretations
✓ Ask students to interpret what they observed during any aspects of this exercise.

● Conclusions
✓ Have students articulate any conclusions they reached as a result of this exercise
experience.

● Applications
✓ Explore with students how they might apply what they learned in this exercise to
situations at their workplace.

3-68 Distributed Programming With Java Technology

Object Activation
Before the release of the Java 2 SDK, a remote object had to be
instantiated and running on the server before it could be used. This
can be inefficient, because not all objects are used all the time. For
example, a remote object that is used only on Sundays should not be
required to be running seven days per week. The Java 2 SDK changes
this. There is a new remote object class, which complements
UnicastRemoteObject: java.rmi.activation.Activatable. An
object of this type can be activated (instantiated), deactivated
(potentially serialized), and reactivated (deserialized) on demand. All
this is completely transparent to the client: as far as the client is
concerned, all objects are constantly active, just like before.

Remote Method Invocation (RMI) 3-69

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
Before activation, a server was required to instantiate and register the
remote object (servant). With activation, the remote object is no longer
instantiated right away. Instead, information about the remote object is
generated and registered. After doing this, the server can exit. Because
the server is not actually a server anymore (it does not serve remote
requests, it just sets everything up), it is usually called XxxSetup, not
XxxServer. The activation, deactivation, and reactivation of remote
objects is handled by the new RMI daemon, rmid.

The Activatable Version of Echo

To convert the UnicastRemoteObject version of Echo to an

Activatable version, you must change the servant (EchoImpl) class.
The server (EchoServer) in its present form is not needed anymore.
Instead, you need a class to set up the activation system. To indicate
the new role of the server, rename it EchoSetup. The client
(EchoClient) does not have to be changed, because activation is a
server-side issue.

Modifying EchoImpl

To modify EchoImpl, complete the following steps:

1. Make the appropriate imports in the implementation class.

import java.rmi.*;
import java.rmi.activation.*;

2. Modify the class declaration so that the class now extends from
java.rmi.activation.Activatable.
public class EchoImpl
extends Activatable implements Echo {

3. Remove or comment out the old no-argument constructor.

//public EchoImpl throws RemoteException {};

4. Declare a two-argument constructor in the implementation class.

public EchoImpl(ActivationID id, MarshalledObject data)
throws RemoteException {
// Register the object with the activation system
// then export it on an anonymous port
super(id, 0);
}

3-70 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
Modifying EchoServer to EchoSetup

Unlike the RMI server class that must stay alive as long as the
implementation needs to be made available, the job of the “setup”
class is to create all the information necessary for the activatable class,
without creating an instance of the remote object.

The setup class passes information about the activatable class to rmid,
and registers a remote reference (an instance of the activatable class's
stub class) and an identifier with the rmiregistry. The setup class
then exits. There are six steps to creating a setup class:

1. Make the appropriate imports in the setup class.

import java.rmi.*;
import java.rmi.activation.*;
import java.util.Properties;

2. Install a SecurityManager.
System.setSecurityManager(new RMISecurityManager());

3. Create an ActivationGroup instance.

Properties props =
(Properties)System.getProperties().clone();
ActivationGroupDesc agd = new
ActivationGroupDesc(props,null);

ActivationGroupID agi =
ActivationGroup.getSystem().registerGroup(adg);
ActivationGroup.createGroup(agi,agd,0);

Remote Method Invocation (RMI) 3-71

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
4. Create an ActivationDesc instance.

The job of the ActivationDesc is to provide all the information that

rmid requires to create a new instance of the implementation class.
This information consists of several parts:

● The class name passed as a String argument.

● The location of remote code as described in a String, which takes

on the form of a URL (this is not a URL class).

● A MarshalledObject instance. You can use the instance of

MarshalledObject to pass configuration data to the servant.
(Look at the constructor of EchoImpl to see how this is
accomplished.) In the example, no special configuration data is
needed, so this argument is going to be null.

This is all that is needed to create an ActivationDesc. The following

is the code segment that creates this object:
String location = new String(“http://localhost:8080/”);
MarshalledObject data = null;

ActivationDesc desc =
new ActivationDesc("EchoApp.EchoImpl", location, data);

5. Remove the reference to the implementation class creation,

declare an instance of your remote interface, and register the
activation descriptor with rmid.
Echo echoRef = (Echo)Activatable.register(desc);

6. Bind the stub that was returned by the Activatable.register

method to a name in the rmiregistry.
Naming.rebind("Echo", echoRef);

3-72 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
Running the Activatable Version of Echo

To run the activatable version of Echo, complete the following steps:

1. Start the rmiregistry. Ensure that the registry is started with no

class path or that the class path does not include a path to any of
the classes the client will download.

2. Start the activation daemon, rmid.

3. Run the setup program.

4. Run the client program.

Remote Method Invocation (RMI) 3-73

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
Exercise: Object Activation

Exercise objective - Modify the server side of the Echo application to

activate the Echo servant on demand, instead of at server startup.

Tasks

Modify the Echo Application (EchoServer and EchoImpl)

Complete the following steps:

1. Change the directory to labfiles/mod3-rmi/lab/activate.

2. In this directory, you will find the well-known Echo application.

Change EchoImpl to be activatable, and change EchoServer to
EchoSetup.

Start the Modified Echo Application

1. Start the activation daemon, rmid.

2. Run the ClassFileServer.

3. Run the setup program.

4. Start the client.

✓ See the README file in /labfiles/mod3-rmi/solutions/activate, which lists the
necessary commands in greater detail.

3-74 Distributed Programming With Java Technology

● Experiences
✓ Ask students what their overall experiences with this exercise have been. You may want to
go over any trouble spots or especially confusing areas at this time.

● Interpretations
✓ Ask students to interpret what they observed during any aspects of this exercise.

● Conclusions
✓ Have students articulate any conclusions they reached as a result of this exercise
experience.

● Applications
✓ Explore with students how they might apply what they learned in this exercise to
situations at their workplace.

Remote Method Invocation (RMI) 3-75

Objects as Parameters in Remote Calls

One of the unique features of RMI is that it can pass full Java objects to
and from the server. Because RMI is 100 percent Java-to-Java, RMI can
pass objects that are subtypes (or subclasses) of a declared parameter
or return type. This allows RMI client applications to pass objects to
the server that have not been previously defined.

RMI “Agents”
By exploiting this capability of leaving the actual implementation of an
object until runtime, you can design an RMI system in which the
server receives objects from clients and executes those objects on the
client’s behalf.

This approach to dynamically executing an object passed from one

address space to another is known as the “command” design pattern.

3-76 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
To make an agent known to the server, declare something. A Java
interface named Agent.java looks like the following:
1 package AgentApp;
2
3 public interface Agent {
4 void run();
5 }

This interface describes a class that can do “something.”

To support the environment of the Agent, you must develop, in the

usual way, an RMI system. First, examine the interface of the remote
object (the servant), which is responsible for accepting the Agent,
calculating it, and sending it back:
1 package WorkerApp;
2 import java.rmi.*;
3 import AgentApp.Agent;
4
5 public interface Worker extends Remote {
6 Agent accept (Agent agent) throws RemoteException;
7 }

Not surprisingly, you will have a WorkerServer, a WorkerImpl, and a

WorkerClient together with the worker interface. In addition, you
will have one example of a class that implements the Agent interface:
CalcFactorial.

The WorkerServer is not listed here. It is analogous to the

EchoServer you already know. The WorkerImpl is a little different; it
must trigger the calculation in the Agent it receives. The following is
the code segment that deals with the Agent:
1 public Agent accept (Agent agent) {
2 agent.run();
3 return agent;
4 }

The servant calls the Agent’s run method and passes the modified
Agent back as the return argument.

The WorkerClient method creates a new Agent (a CalcFactorial,

which implements the Agent interface), and then calls the accept
method of the servant.

Remote Method Invocation (RMI) 3-77

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
1 CalcFactorial agent = new CalcFactorial(100);
2 agent = (CalcFactorial)workerRef.accept(agent);
3 System.out.println("Got the following
result: " + agent.getResult());

The following is an example of the CalcFactorial class:

1 package AgentApp;
2
3 public class CalcFactorial implements Agent,
java.io.Serializable {
4 private int value;
5 private double result;
6
7 public CalcFactorial() { }
8
9 public CalcFactorial(int value)
{ this.value = value; }
10
11 public void run()
12 {
13 result = 1;
14 for (int i=1; i<=value; i++)
15 {
16 result *= i;
17 }
18 //The text appears on the virtual
19 //machine the agent is really
20 //calculating on.
21 System.out.println("\nCalculated the value "
+ result + " on this (virtual) machine!\n");
22 }
23
24 public double getResult () {
25 return result;
26 }
27 }

As the name indicates, the CalcFactorial class calculates the

factorial of an integer and passes the result back in a value of type
double.

3-78 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
The following are the major steps that occur when you start up
everything and run the client:

1. A new CalcFactorial class is created on the client.

2. The CalcFactorial class travels over the wire to the server (it is
serialized on the client, sent to the server, and deserialized on
arrival). The server is now in possession of an exact copy of the
CalcFactorial class the client created.

3. On the server’s copy of the CalcFactorial class, the method run

is called. This call changes the state of the server’s
CalcFactorial: Its private result field is updated. The
CalcFactorial on the server is no longer equal to the
CalcFactorial on the client.

4. The modified CalcFactorial is sent back to the client, as the

return argument. Again, the object is serialized, sent over the
wire, and deserialized on the client. On the client, throw away
the old CalcFactorial, assign the new one to the variable
agent, and then print its result.

What is the outcome of this setup? You can to move a compute-

intensive task to the server, and get the result back to the client.

You might wonder if, instead, it would have been simpler to make the
Agent a remote interface, and create an AgentImpl class. But this
method’s setup is powerful enough to create additional classes that
implement the Agent interface. Because a client takes the name of the
Agent from the command line, you do not have to change anything to
support the new agents. You do not have to recompile any of the
Worker classes, stop and restart the RMI system, or install modified
code on the server. The name Worker is appropriate; it is a generic
service to “work” on compute-intensive tasks.

Remote Method Invocation (RMI) 3-79

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
Exercise: Objects as Parameters in Remote Calls

Exercise objective – Use the provided Agent code and test whether
anything can be gained by letting the Agent run on somebody else’s
machine.

Preparation
Team up with somebody else (preferably somebody with either a
slower or a faster machine than you have).

Tasks

Compile and Run the Provided Agent Application

Complete the following steps:

1. Change the directory to labfiles/mod3-rmi/agent.

2. Compile and run the provided application locally.

✓ Do not forget to start the httpsrv.ClassFileServer.
✓ The directory labfiles/mod3-rmi/agent contains a README file with more explicit
instructions and troubleshooting information.

3-80 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
Run the Agent on Another Machine

Complete the following steps:

1. Let your agent run on the machine of your classmate.

2. Modify your client to produce an indication of the time spent

sending the Agent off and let it calculate (use
System.currentTimeMillis to produce a timestamp). You
probably have to modify your agent, or all you measure is the
network delay (the calculation of the factorial is too fast). For
example, you can repeat the calculation a couple of times in a
secondary loop, to increase the time the Agent spends on the
remote system.

Can you see a performance gain by sending the Agent to

somebody else’s machine?
✓ The calculation had to be repeated 100,000 times to get a measurable result.Think About
the Agent Interface

3. Think about the Agent interface for a minute. Is it a good

abstraction?
✓ It depends on how you look at it. If all you care about is the server, then the interface is
okay. It describes everything a server must know about an Agent. If you want to build a
universal client as well, the interface is not good enough. Right now, the client has to
“know” the intimate details of CalcFactorial to get at its result. A universal client would
rely only to an interface as well. An expanded Agent interface would include methods for
data retrieval, such as String getResultAsText, Class getResult, or maybe even a callback
method such as void paintResult(Graphics gfx, Rectangle rect). An even better
solution would be to develop two interfaces: one for usage on the server, and one for
usage on the client. An Agent implementation would then have to implement both
interfaces.

This exercise was about a “compute” server. Are there other

possibilities for generic server types that accept objects of which
they know only the interface?
✓ This goes back to the generic agent discussion of Module 1.

Remote Method Invocation (RMI) 3-81

● Experiences
✓ Ask students what their overall experiences with this exercise have been. You may want to
go over any trouble spots or especially confusing areas at this time.

● Interpretations
✓ Ask students to interpret what they observed during any aspects of this exercise.

● Conclusions
✓ Have students articulate any conclusions they reached as a result of this exercise
experience.

● Applications
✓ Explore with students how they might apply what they learned in this exercise to
situations at their workplace.

3-82 Distributed Programming With Java Technology

HTTP Tunneling
RMI provides a means for clients behind firewalls to communicate
with remote servers. This allows you to use RMI to deploy clients on
the Internet, such as in applets available on the World Wide Web.

Remote Method Invocation (RMI) 3-83

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
Traversing the client’s firewall can slow down communication, so RMI
uses the fastest successful technique to connect between client and
server. The technique is discovered by the reference to
UnicastRemoteObject on the first attempt the client makes to
communicate with the server. It tries each of the following three
possibilities:

● Communicate directly to the server’s port using sockets. If this

fails, build a URL to the server’s host and port and use an HTTP
POST request on that URL, sending the information to the skeleton
as the body of the POST. If successful, the results of the post are
the skeleton's response to the stub.

● If this also fails, build a URL to the server’s host using port 80 (the
standard HTTP port) using a CGI script that forwards the posted
RMI request to the server.
✓ The client application may disable the packaging of RMI calls as HTTP requests by setting
the java.rmi.server.disableHTTP property to be true.
✓ By default only port 80 is used.

3-84 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
Exercise: Developing an RMI Application From Scratch
(Optional)

Exercise objective – Develop an entire RMI application from scratch.

No specific templates are provided, but you can use the Echo files as a
guideline. Only the interface of the remote object is provided.

Tasks

Implement and Run a Remote Counter Application

Take the Echo files as templates to build a complete RMI application

based on the interface named RemoteCount.java. The client will call a
remote method a given number of times and measure the time
between invoking a call and getting back an answer. The average time
is then displayed on the console.

1. Change the directory to labfiles/mod3-rmi/remotecount.

2. Implement the server and the servant, based on the interface

RemoteCount.java. The methods in the interface have the
following purposes:

▼ sum returns an int value showing how many times the remote
method increment was invoked. It should be called after the
iterations have finished.

▼ setZero does not return any value. It sets the increment

variable back to 0 and should be called before a new series of
loops start.

▼ increment increments a counter on the server by 1.

Remote Method Invocation (RMI) 3-85

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
3. Develop a client that you can start with a command, such as:
java Count.RemoteCountClient localhost 1000

This calls the remote method 1000 times. The client console should
then look like the following:
sum = 0
Now calling the remote method
Average Time for a method call = 1.7 milliseconds

The client measures the time at the beginning of the loop and at
the end. Dividing the time measured by the number of loops
returns the average time a remote call takes. The method for
getting the current time is System.currentTimeMillis.

4. Team up with your neighbor to make the remote calls on another

machine and to see if there are differences in time.

3-86 Distributed Programming With Java Technology

● Experiences
✓ Ask students what their overall experiences with this exercise have been. You may want to
go over any trouble spots or especially confusing areas at this time.

● Interpretations
✓ Ask students to interpret what they observed during any aspects of this exercise.

● Conclusions
✓ Have students articulate any conclusions they reached as a result of this exercise
experience.

● Applications
✓ Explore with students how they might apply what they learned in this exercise to
situations at their workplace.

Remote Method Invocation (RMI) 3-87

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
3
Check Your Progress

Before continuing on to the next module, check that you were able to
accomplish the following in this module:

❑ Describe the RMI architecture, its layers and garbage collection

❑ Implement an RMI server and client in the Java programming

language

❑ Generate client stubs and skeletons for remote services using the
RMI stub compiler

❑ Define the RMI registry and describe how it works

❑ Explain the security issues related to RMI

3-88 Distributed Programming With Java Technology

What applications do you have that could take advantage of RMI?

Remote Method Invocation (RMI) 3-89

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
Java Interface Definition Language
(JavaIDL) 4

Objectives
Upon completion of this module, you should be able to:

● Describe the basic CORBA object management architecture

● Describe the role of JavaIDL in relation to other commercial Java

CORBA products

● Create and deploy a JavaIDL server object and a JavaIDL client

application

● Describe how the JavaIDL bootstrapping process works

● Describe how IDL is mapped to the Java programming language

● Explain why the RMI Agent example does not work with the
current CORBA

JavaIDL adds CORBA (Common Object Request Broker Architecture)

capability to the Java platform, providing standards-based
interoperability and connectivity. Runtime components include a fully
compliant Java ORB for distributed computing using Internet inter-
operability protocol (IIOP) communication.

4-1
Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
Relevance

Discussion – Consider the following questions:

● How would you wrap existing legacy code (C, C++, COBOL)?

✓ This is the primary function of CORBA. This technology is exemplified by JavaIDL. CORBA
allows problems to be modeled in a single language (IDL) and then compiled to other
languages.

● How does a system publish its services in such a way that any
client can request any service?
✓ This is the job of the ORB—to provide a central naming service for all services.
✓ This module covers JavaIDL. All of the IDL mapping and features described are supported
by JavaIDL. There are other features that are supported by other implementations of
CORBA—this document is intended only to introduce students to JavaIDL.
✓ JavaIDL supports IIOP, a generic ORB core that communicates via IIOP; an idltojava
“compiler” that generates both stub and skeleton code; and an implementation of the
CORBA Object Service (COS) Naming service called tnameserv.

4-2 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
Additional Resources

Additional resources – The following resources can provide

additional detail on the topics presented in this module:

● The Object Management Group. [Online]. Available:

http://www.omg.org/

● Mapping of OMG IDL to Java. [Online]. Object Management Group.

Available: ftp://www.omg.org/pub/docs/formal/98-02-
29.pdf

● ORB Portability Java Language Mapping. [Online]. Object

Management Group. Available:
ftp://ftp.omg.org/pub/docs/orbos/97-11-12.pdf

● Revised ORB Portability IDL/Java Language Mapping. [Online].

Object Management Group. Available:
ftp://ftp.omg.org/pub/docs/orbos/98-01-06.pdf

● Java to IDL Mapping. [Online]. Object Management Group.

Available: ftp://ftp.omg.org/pub/docs/orbos/98-02-01.pdf
ftp://ftp.omg.org/pub/docs/orbos/98-03-08.pdf (errata to
the above document)

● Objects By Value. [Online]. Object Management Group. Available:

ftp://ftp.omg.org/pub/docs/orbos/98-01-01.pdf

● The Portable Object Adaptor. [Online]. Object Management Group.

Part of the complete CORBA 2.2 specification. Available:
ftp://www.omg.org/pub/docs/formal/98-02-14.pdf

● RMI over IIOP: JavaOne session slides. [Online]. Sun

Microsystems, Inc. Available:
http://java.sun.com/javaone/javaone98/sessions/T406
/kgh1.htm

● Orafali, Robert and Dan Harkey. Client/Server Programming with

Java and CORBA Second Edition. 1998. Wiley Computer Publishing.

● Lewis, Geoff, Steven Barber, and Ellen Siegel. Programming with

JavaIDL. 1997. Wiley Computer Publishing.

Java Interface Definition Language (JavaIDL) 4-3

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
Module Overview

CORBA enables you to bridge the gaps among different programming

languages, hardware architectures, operating systems, and locations.
In this module, you will look at how to build a Java CORBA server
object, and a Java CORBA client application that talks to the server
object, using JavaIDL. In theory, the client application could also talk
to an equivalent C++ CORBA server object, and the server object could
talk to the client application. However, doing that is not included in
this course.

This module includes a brief look at the core of the Object

Management Architecture (OMA): the Object Request Broker (ORB),
and JavaIDL.

4-4 Distributed Programming With Java Technology

Object Request Broker

The ORB is the core of the reference model. It provides the

communications infrastructure that enables objects to transparently
make and receive requests and responses in a distributed
environment. It is the foundation for building applications from
distributed objects and for achieving interoperability between
applications in heterogeneous environments.

The ORB forms a bridge that eliminates differences in location,

platform, and programming language. ORBs use a standard
communications protocol specified by the General Interoperability
Protocol (GIOP). However, this course does not define the transport
mechanism, so the OMG also created a specification for
communications over TCP/IP, the Internet Inter-ORB Protocol (IIOP).

Java Interface Definition Language (JavaIDL) 4-5

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
Consistent with its focus on interfaces, CORBA does not specify how
the ORB is to be implemented; it can be a daemon process, library, or
combination thereof.

Note – The actual degree of involvement of the ORB in service

requests and responses is not specified by the standard. In some
implementations the ORB is involved only in the initial establishment
of the communications link and not in subsequent method
invocations.

4-6 Distributed Programming With Java Technology

ORB Implementation

Static and Dynamic Invocation

CORBA defines two mechanisms for invoking object methods: static
invocation and dynamic invocation. Static invocation is limited to the
interfaces known when the client code is compiled. Dynamic
invocation allows the client to discover interfaces at runtime by
interrogating the ORB. This is a powerful feature and is vital for
building flexible systems that can respond to frequent changes in the
real world.

Java Interface Definition Language (JavaIDL) 4-7

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
Interface Repository
The key feature that makes dynamic invocation possible is the
interface repository. In its simplest form, this is a database of interfaces
populated by the IDL compiler. Architecturally speaking, it is one of
the most critical components in CORBA, as it contains most of the
“metadata” for the whole federation of objects. It is also involved in
type-checking of method invocations, as well as referencing interfaces
and methods across ORBs.
✓ Dynamic invocation relies on the use of the Any type. These parameters are type-checked
at runtime by interrogating the interface repository. The current JavaIDL does not provide
support for an interface repository, and the Any type is beyond the scope of this course.

Object Adapter
In general, an adapter allows objects with incompatible interfaces to
communicate. CORBA defines the object adapter as an ORB
component that provides object reference, activation, and state-related
services to an object implementation. These include:

● Registering implementation objects with the ORB

● Interpreting and translating object references

● Locating implementation objects

● Activating and deactivating implementation objects

● Invoking methods
✓ Explain that the object adapter, represents a basic design pattern of OO software
engineering. Like an electrical adapter, it is used to connect interfaces which do not
communicate directly.

To perform these tasks, the object adapter maintains an

implementation repository for storing information that describes
object implementations.

While an ORB must have, at a minimum, the basic object adapter

(BOA) specifically defined by CORBA, it can also have special-purpose
object adapters. You can use a specialized object adapter, for example,
to implement persistence.

4-8 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
CORBAservices
CORBAservices support basic functions for using and implementing
distributed applications and are independent of application domains.
For example, the Lifecycle Service defines interfaces for creating,
deleting, copying, and moving objects; it does not dictate how the
objects are implemented in an application. Refer to the OMG
document CORBAservices for discussion of the services defined by
CORBA.

CORBAfacilities
This is a collection of shared services at a higher architectural level
than the CORBAservices.

In general:

● Horizontal facilities can be shared across many applications, such

as printing and electronic mail.

● Vertical facilities apply to a specific business domain, such as

finance or manufacturing.

A number of facilities are defined by OMG (see the document

CORBAfacilities).

Java Interface Definition Language (JavaIDL) 4-9

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
Horizontal Facilities

The following lists the horizontal, or generic, facility categories:

● User interface facilities include Compound Presentation, Desktop

Management, Rendering Management, Scripting, and User
Support Facilities.

● Information management facilities include Compound

Interchange, Data Encoding and Representation, Data Interchange,
Information Exchange, Information Modeling, Information Storage
and Retrieval, and Time Operations Facilities.

● System management facilities include Collection

Management, Consistency, Customization, Data Collection, Event
Management, Instance Management, Instrumentation, Policy
Management, Process Launch, Quality of Service Management,
Scheduling Management, and Security Facilities.

● Task management facilities include Agent, Automation, Rule

Management, and Workflow Facilities.

4-10 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
Vertical Facilities

The following lists the vertical CORBAfacilities that are currently

identified by OMG in the following areas:

● Accounting

● Application development

● Computer-integrated manufacturing

● Currency

● Distributed simulation

● Information superhighways

● Internationalization

● Mapping

● Oil and gas exploration and production

● Security

● Telecommunications

Java Interface Definition Language (JavaIDL) 4-11

Wrapping Legacy Code With CORBA

One of the primary features of CORBA is the support for multiple

programming languages across multiple platforms. CORBA vendors
have had time to create IDL compilers for a variety of platforms and
languages, and have sold a number of solutions for “wrapping” a
large body of existing code with a CORBA implementation.

● Legacy code is modeled using IDL. This file represents the

“contract” of the services the server implementation (legacy code)
provides. Some compromises might have to be made to make the
IDL work.

● On the server side, the IDL is compiled and “linked” into the
legacy code. This can happen by calling legacy libraries, or, in
some cases, compiling the legacy code directly into the skeletons
created by the IDL compiler.

4-12 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
● On the client side, the IDL is compiled to the programming
language desired, such as Java, and the client application can
make calls to the methods that represent the legacy “services.”

● On both sides, the communication between the skeleton code and

stub code and the ORB is transparent. The ORB code is designed
to insulate the client/server from the specifics of the object
implementation.

Java Interface Definition Language (JavaIDL) 4-13

JavaIDL in Relation to CORBA

JavaIDL – A Full ORB?

JavaIDL is an Object Request Broker (ORB) provided with the Java 2
SDK. You can use it to define, implement, and access CORBA objects
from the Java programming language. JavaIDL is compliant with the
CORBA/IIOP 2.0 specification and the IDL-to-Java programming
language mapping.

The JavaIDL ORB supports transient CORBA objects (objects whose

lifetimes are limited by their server process’s lifetime). JavaIDL also
provides a transient nameserver to organize objects into a tree-
directory structure. The nameserver is compliant with the Naming
Service Specification described in CORBAservices.

4-14 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
Is a Commercial ORB Necessary?
At first glance, JavaIDL seems to be a complete ORB, together with an
implementation of the OMG Naming Service. The obvious question is
therefore: Is JavaIDL all that is necessary or do you still need a
commercial ORB? The answer is not simple. You need to compare
JavaIDL with the current Java programming language-compliant
commercial ORBs. The following paragraphs look at JavaIDL, both the
client and the server parts.

JavaIDL on the Client

JavaIDL is a good choice on the client, as long as you do not need any
special features: tunnelling of IIOP through HTTP (to circumvent
firewalls), and using SSL to encrypt the IIOP datastream. If you need
these features, you must switch to another ORB on the client. RMI
supports both features (SSL since Java 2 SDK), so there is a chance that
JavaIDL will support tunnelling and SSL as well.

JavaIDL on the Server

For developing, and many simple purposes, JavaIDL can be a good

choice on the server. However, the list of unsupported features is
significantly longer on the server than it was on the client. It includes
the following:

● Objects written in a programming language other than Java

● An interface repository

● Portable object adapter (POA)

● Transparent object activation

● Any CORBAservice besides the naming service

● A persistent naming service

● A more robust server, which can gracefully recover from a crash

(by periodically persisting the state of all servants running, and by
supporting transactions)

Java Interface Definition Language (JavaIDL) 4-15

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
In a nutshell, you should use a commercial ORB on the server for
production. On the client, JavaIDL is good enough most of the time,
and has the additional advantage of being already installed (if the
clients are on Java 1.2.)
✓ An experienced programmer who does not use the Java programming language and is a
CORBA developer might ask if one can deploy an applet that was developed using
Visibroker onto a JavaIDL client. With C++, this is not possible: an application that is
intended to run on a Orbix runtime cannot be delivered with Visibroker stubs. With the
Java programming language, the same thing is usually possible. A “binary compatible
stub” standard was agreed on. You will only encounter problems if your applet uses
services that are not available on the client runtime; for example, the COS naming service
of JavaIDL is not available on the Netscape Communicator built-in CORBA runtime. For
such a case, it is possible to download and replace the ORB itself at runtime, using an
applet parameter.

4-16 Distributed Programming With Java Technology

Interface Definition Language Basics

After learning the basics, you can start with the first step of creating
your own CORBA application: declaring the interface of your server
object. With RMI, you used the Java programming language to do this;
you declared a Java interface that described the server object.

CORBA uses a specially designed language to accomplish the same

task: Interface Definition Language (IDL). A compiler (idltojava)
later translates IDL-to-Java code, generating (among other items) a
Java interface that you can use on the client in much the same way you
used the Java interface with RMI.

The IDL standard, like a Java interface, declares only the interface of
the server objects, abstracting all of the implementation details.

Java Interface Definition Language (JavaIDL) 4-17

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
For example:
1 module EchoApp {
2 interface Echo {
3 string sayEcho(in string myName);
4 };
5 };

This interface definition is passed to the IDL compiler, which generates

the following:

● Echo.java – This is the Java interface mirroring the IDL interface

using the Java programming language syntax.

● EchoHelper.java – This object is primarily used to cast a generic

org.omg.CORBA.Object to the Echo type. (You will see later why
you cannot cast the usual way.)

● EchoHolder.java – This object is used if an Echo object is being

passed as an out or inout parameter in a method call. You will
see more about holders later on.

● _EchoImplBase.java – This object is extended by the actual Echo

implementation. _EchoImplBase.java provides code for the
server side to manage interaction between the ORB and the actual
Echo object.

● _EchoStub.java – This object is implicitly used by the client. It

implements the interface Echo. It provides code for the client to
allow it to invoke the method sayEcho using local Java
programming language semantics.

All of the networking code required for making and responding to

requests is automatically generated by the IDL compiler, and the client
and server code can be independently targeted for any language for
which the ORB vendor supplies an IDL mapping.

4-18 Distributed Programming With Java Technology

IDL Language Mappings

Currently there are several language mappings available and more
will be added when the need arises. The OMG has standard mappings
for the following programming languages: C, C++, Smalltalk, COBOL,
Ada ’95, and Java. Most of the existing commercial ORBs support
generation of both client and server code.

Java Interface Definition Language (JavaIDL) 4-19

Important IDL Keywords

So that you can start coding soon, this section includes only the most
important IDL keywords. More IDL details are covered after you have
deployed your first JavaIDL application.

The following is the simple object again:

1 module EchoApp {
2 interface Echo {
3 string sayEcho(in string myName);
4 };
5 };

4-20 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
Look at these keywords:

● module – This keyword describes a naming context, in much the

same way a Java programming language package (“Java
package”) does. The idltojava compiler translates a module to a
Java package.

● interface – This keyword describes the interface of a server

object. It maps directly to a Java interface with the same name.

● string – IDL, like the Java programming language, has a range of

primitive datatypes. In IDL, string is one of them (unlike Java,
where String is a class). Nonetheless, an IDL string maps to a Java
String.

● in – This keyword denotes that the parameter myName travels one

way from the client to the server.

● string sayEcho(in string myName) – While not a keyword,

this whole sentence is the declaration of a method named
sayEcho. This declaration is mapped to a Java method with the
same name. Because string translates to String, the full Java
programming language method (“Java method”) declaration is
String sayEcho(String myName).

This quick IDL introduction is enough to enable you to create a

JavaIDL server and client.

Java Interface Definition Language (JavaIDL) 4-21

JavaIDL Architecture Overview

JavaSoft’s JavaIDL generates both stub and skeleton code from an IDL
file by compiling the IDL file with idltojava. The IDL compiler for
JavaIDL generates a set of stubs and skeletons that provide
functionality for static invocation on CORBA objects.

The following describes the operation of a client/server CORBA

system:

● On the client side, the client invokes a method on an object that is

“stubbed” to appear local to the client. The stub tells the client’s
ORB what arguments to marshal and tells the ORB to invoke the
CORBA object identified by the remote reference. The ORB
“knows” how to marshal arguments, but not which arguments to
marshal for a particular method.

● The ORB uses the object reference to determine which remote ORB
should receive the request, passes the marshalled parameters over
the wire to the ORB, and waits for a result.

4-22 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
● On the server side, the ORB receives the request, inspects the
object reference to determine which server contains the object, and
invokes the skeleton for that object. The skeleton tells the ORB
how to unmarshal the parameters, and invokes the server
implementation of the method requested.

● The server method returns a result to the skeleton, which arranges

how to marshal the result with the server-side ORB and returns it
to the stub. The stub contacts its ORB to unmarshal the result and
return it to the client.

Java Interface Definition Language (JavaIDL) 4-23

Creating and Deploying a JavaIDL Application

Creating a JavaIDL Application

There are six major steps involved in creating the JavaIDL application
(client and server) from the IDL specification:

1. Develop or acquire the IDL.

2. Compile the IDL using the JavaIDL idltojava compiler.

3. Create the implementation class (the servant) for the interface

defined in the IDL.

4. Create the server, which manages the servant instances.

5. Create the client, which uses the remote object (in the end, the
client uses the servant).

6. Create the Java class files. Use the Java technology compiler (“Java
compiler”) javac to create the Java technology class (“Java class”)
files for all of the Java technology files (“Java files”).

4-24 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
Step 1: Develop or Acquire the IDL

The IDL file is the contract between the server object and the client
application. It is usually developed at the time a new server object is
developed. Therefore, in a productive CORBA environment, the IDL
for existing server objects can usually be acquired, if you have to
develop a matching Java client.

Step 2: Compiling the IDL

Assuming the JavaIDL directory is in your path, compile the IDL file
by specifying what you want idltojava to do with the file. For
example, the simplest use of the compiler is to generate both client
code (stubs) and server code (skeletons), and not to use a preprocessor
first (CPP).
idltojava -fno-cpp Echo.IDL

By default, idltojava generates the following Java files, in the

EchoApp package:
_EchoImplBase.java
_EchoStub.java
Echo.java
EchoHelper.java
EchoHolder.java

Step 3: Create the Implementation Class (Servant)

For each interface, you must write a servant class. Instances of the
servant class implement ORB objects. Each instance implements a
single ORB object, and each ORB object is implemented by a single
servant.

idltojava generates a base class _<interface_name>ImplBase for

every IDL interface declared in the IDL file. This base class must be
extended to create the servant.

The servant code therefore extends the _EchoImplBase class, and

provides an implementation for all of the methods declared in the IDL
file. In this case this is just one: sayEcho.

Java Interface Definition Language (JavaIDL) 4-25

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
Despite the fact that the implementation classes are called servants, the
usual naming convention is to append Impl to the Java interface name:
EchoImpl.java.
1 package EchoApp;
2
3 class EchoImpl extends _EchoImplBase {
4 public String sayEcho(String myName) {
5 return "\nHello " + myName + "!!\n";
6 }
7 }

Step 4: Create the Server

The server is what the ORB uses to instantiate and publish the servant.
In its simplest form, the server pre-instantiates all the needed servants
and keeps them alive forever (or at least as long as the server lives). In
a production system this is usually not good enough. To keep
thousands of servants alive, even when they are not actively used by
clients, is a waste of resources. In this case, more complex servers must
be developed and used. They must be able to instantiate servants on
demand, and maybe even persist and discard them after a period of
non-usage. The server, EchoServer.java, is a simple one:
1 package EchoApp;
2
3 import org.omg.CosNaming.*;
4 import org.omg.CORBA.*;
5
6
7 public class EchoServer {
8 public static void main(String args[]) {
9 try{
10 // create and initialize the ORB
11 // pass the command line arguments to it
12 ORB orb = ORB.init(args, null);
13
14 // create servant and register it with the ORB
15 // it is now a CORBA object,
16 // but not yet retrievable
17 // via the COS name server.
18 EchoImpl echoRef = new EchoImpl();
19 orb.connect(echoRef);
20
21 // get the remote reference to the
22 // COS name server

4-26 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
23 // Narrow it to the correct type.
24 org.omg.CORBA.Object objRef =
orb.resolve_initial_references("NameService");
25 NamingContext ncRef =
NamingContextHelper.narrow(objRef);
26
27 // bind the object reference
28 // to the name "Echo"
29 NameComponent nc =
new NameComponent("Echo", "");
30 NameComponent path[] = {nc};
31 ncRef.rebind(path, echoRef);
32 System.out.println("Echo object
ready and bound to the name 'Echo'!");
33 System.out.println("It's object reference
is: " + orb.object_to_string(echoRef));
34
35 // now you have to stop, or the virtual machine
36 // will exit, killing with it the newly
37 // created Servant object.
38 // Stop in a non CPU intensive way
// by waiting forever
39 java.lang.Object sync =
new java.lang.Object();
40 synchronized (sync) {
41 sync.wait();
42 }
43
44 } catch (Exception e) {
45 System.err.println("ERROR: " + e);
46 e.printStackTrace(System.out);
47 }
48 }
49 }

Java Interface Definition Language (JavaIDL) 4-27

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
As you can see, you not only have to instantiate the servant and
register it with the ORB, but also register it with the name server. After
the servant is registered, the client application can retrieve and use the
servant’s object reference. You will see how this works in the following
section.
✓ Experienced CORBA developers might ask why this example does not have to use an
Object adapter (BOA or POA). With a C++ server, you would have been forced to use one.
The answer lies in the new IDL to Java technology binding standard: it declares in the
ORB interface two new methods: connect and disconnect. These two are used to connect
and disconnect a transient servant to the ORB, without an object adapter in between.
✓ This is different from RMI. You have to explicitly wait at the end of the main thread to avoid
exiting the virtual machine. While the source code of RMI and JavaIDL has not been
looked at, it is believed that the difference must be in the way the two systems use
threads. RMI uses “regular” threads, while JavaIDL uses “daemon” threads. A Java virtual
machine exits as soon as the last non-daemon thread finishes.

Step 5: Create the Client Application

The following is an example of how to create a client application.

1 package EchoApp;
2 import org.omg.CosNaming.*;
3 import org.omg.CORBA.*;
4
5 public class EchoClient {
6 public static void main(String args[]) {
7 try {
8 // create and initialize the ORB
9 // pass the command line arguments to it
10 ORB orb = ORB.init(args, null);
11
12 // get the remote reference to the COS name
13 // server. Narrow it to the correct type.
14 org.omg.CORBA.Object objRef =
orb.resolve_initial_references("NameService");
15 NamingContext ncRef =
NamingContextHelper.narrow(objRef);
16
17 // resolve the Object Reference in Naming
18 NameComponent nc = new NameComponent
("Echo", "");
19 NameComponent path[] = {nc};
20 org.omg.CORBA.Object tempEchoRef =
ncRef.resolve(path);
21 // we have an org.omg.CORBA.Object,
22 // but we need an Echo

4-28 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
23 Echo echoRef = EchoHelper.narrow(tempEchoRef);
24
25 // call the Echo server object and print results
26 String reply = echoRef.sayEcho("Martin");
27 System.out.println(reply);
28
29 } catch (Exception e) {
30 System.out.println("ERROR : " + e) ;
31 e.printStackTrace(System.out);
32 }
33 }
34 }

Echo was used as a name to bind the servant into the name server.
Now use the same name to look it up. The result is an object of type
org.omg.CORBA.Object, but you need an object of type Echo. A
simple cast, such as(Echo)ncRef.resolve(path) does not always
work; it depends on the ORB. The CORBA way to do casting is to use
the narrow-method of the generated helper class
EchoHelper.narrow.

Step 6: Create the Java Class Files

To compile the files, type the following:

javac -d . *.java

The files in the EchoApp packages are compiled automatically, because

both EchoServer and EchoClient are dependent on them.

Java Interface Definition Language (JavaIDL) 4-29

Deploying a JavaIDL Application

First Try: Start Everything Local and Accessible

Assuming there are no compilation problems, you start the name

server first, then the server, and finally the client.
tnameserv

The default port is 900, which can be problematic. The UNIX®

environment hosts do not let you connect to ports below 1024 if you
are not root. Fortunately, you can change the port. Use tnameserv -
ORBInitialPort 1050 to connect using port 1050, for example.

You start the server next, which creates the servant and registers it
with the name server.
java EchoApp.EchoServer

Again, add -ORBInitialPort 1050 as argument to connect to port

1050.

4-30 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
In the last step, start the client.
java EchoApp.EchoClient

-ORBInitialPort 1050 uses port 1050 to connect to the server ORB

and name server.

Second Try: Connect The Client to a Remote ORB

To connect your client to a different machine, add a parameter on

startup.
java EchoApp.EchoClient -ORBInitialHost yourhost

This line connects you to host yourhost. -ORBInitialPort and

-ORBInitialHost can be combined.

This might be a good time to find out why passing arguments works:
nothing was done in the code to support this. You should be able to
find this out by yourself, by looking through your EchoClient source
code and the Java documentation.
✓ The line ORB orb = ORB.init(args, null); passes all the command line arguments
unchanged to the ORB. Unfortunately, the documentation of the valid parameters is a little
hard to find: The ORB.init method documentation reveals nothing. Currently, the
description of the valid parameters can be found in the Java 2 SDK documentation,
JavaIDL part, “Programming Guide”, section “initialization”. Part of this section is also a
description on how to set the parameters for an Applet.

In Module 3, ‘‘Remote Method Invocation (RMI),” you were able to

download stubs on demand, by using the RMIClassLoader on the
client and an HTTP-based code server. Because there is no
IDLClassLoader available, the same thing is not easily possible with
JavaIDL. Usually, the stubs are packaged together with the rest of the
client code and the generated helper classes, before the whole package
is deployed onto the client.

Java Interface Definition Language (JavaIDL) 4-31

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
Exercise: Compiling an Application

Exercise objective – Compile and run the Echo application. In

addition, test the robustness of the JavaIDL system by intentionally
breaking the code of the Echo application.

Tasks

Compile the Echo Application

Work through the six steps described in this module to compile the
Echo application.

1. Change the directory to labfiles/mod4-jidl/lab/echo.

2. Compile the Echo server and client.

Run the Echo Application

Work through the steps described in “Deploying a JavaIDL

Application” to get the Echo application running.

1. Run the Echo application locally on your system.

2. Team up with somebody and distribute the client and the server
part of Echo on two different machines.

Break the Echo Application to Test the Robustness of the JavaIDL

System

After Echo runs successfully, try to break the code in several ways to
test the fault tolerance of the JavaIDL.

1. In the servant code:

▼ Do not implement a method that is in the IDL (or fail to

implement the correct method signature).

▼ Leave out “extends _EchoImplBase”.

4-32 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
2. In the server code:

▼ Comment out orb.connect(echoRef).

▼ Replace NamingContextHelper.narrow(objRef) with an

explicit casting.

▼ Comment out the part at the end, where the server stops
“forever.”

Exercise Summary
Discussion – Take a few minutes to discuss what experiences, issues,
or discoveries you had during the lab exercises.

● Experiences
✓ Ask students what their overall experiences with this exercise have been. You may want to
go over any trouble spots or especially confusing areas at this time.

● Interpretations
✓ Ask students to interpret what they observed during any aspects of this exercise.

● Conclusions
✓ Have students articulate any conclusions they reached as a result of this exercise
experience.

● Applications
✓ Explore with students how they might apply what they learned in this exercise to
situations at their workplace.

Java Interface Definition Language (JavaIDL) 4-33

Bootstrapping the JavaIDL System

Most of the bootstrapping code was shown in the EchoClient and

EchoServer. This section reviews some of the details.

What Is Bootstrapping?
Using remote objects in a client application is easy and transparent,
once you have the references. The bootstrapping protocol describes
how you can get at the initial remote object references in a
standardized way.

If this protocol were not standardized, then every ORB vendor would
do it differently. Consequently, an applet written for Visibroker would
not work if run on a client with only JavaIDL installed. You would
have to download not only the applet, but also the Visibroker ORB to
the client, which wastes bandwidth.

4-34 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
A protocol standardization is in the works. An initial submission was
delivered to OMG in April, 1998. The document is titled
“Interoperable Naming Service” and can be found on
ftp://ftp.omg.org/pub/docs/orbos/98-03-04.pdf. The proposed
standard is not specific to the Java programming language. Other
languages will profit from this standard as well, in the form of
increased source code level portability of CORBA client applications.

JavaIDL conforms to the proposed standard. Because the proposal was

jointly submitted by several OMG members, it is unlikely that the
document will change much before it is adopted.

Bootstrapping the Client Application

Bootstrapping consists of three steps:

1. Obtain a reference to the local ORB.

2. Use the ORB to retrieve the reference to a COS naming service.

3. Use the COS naming service to retrieve the remote object

references.

Alternatively, you can use the ORB to resolve a named object reference
to an actual reference. This might be necessary if no naming service is
available.

Obtain a Reference to the Local ORB

The following line retrieves the ORB, when called from a Java
application:
ORB orb = ORB.init(args, null);

Java Interface Definition Language (JavaIDL) 4-35

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
On startup, the ORB checks the following properties:
org.omg.CORBA.ORBClass
org.omg.CORBA.ORBSingletonClass
org.omg.CORBA.ORBInitialHost
org.omg.CORBA.ORBInitialPort

You can set properties in three places:

● On the command line (this is why ORB.init gets a reference to the

command-line arguments)

● In the form of a Java Properties object (which is passed instead

of null as the second argument in ORB.init)

● As a Java system property

If the property is passed on the command line, you must omit

org.omg.CORBA.

An applet does not have command-line arguments. Instead, an applet

has parameters encoded in the HTML file that loads the applet.
Instead of args, pass a reference to the applet to the ORB:
orb = ORB.init(this, null);

Of the four parameters, only the first two are standardized.

ORBInitialHost and ORBInitialPort are JavaIDL-specific. They are
used to specify the host and IP port number of the machine providing
the initial services (for instance, the naming service).

By setting ORBClass to a specific value, the ORB can be replaced at

runtime. The following applet parameter, for example, will run the
applet using Visibroker 3.0 instead of JavaIDL:
<param name=org.omg.CORBA.ORBClass
value=com.visigenic.vbroker.orb.ORB>

Of course, the specified ORB must be accessible to the applet, either

from the local machine or from the Web server.

4-36 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
Bootstrapping the Client Application

Use the ORB to Retrieve the Reference to a COS Naming Service

To retrieve initial references, the bootstrapping protocol specifies two

calls that must be supported by the ORB:

● ORB.list_initial_services()

● ORB.resolve_initial_references()

The first call delivers an array of strings, describing the services

available on the ORB. Usually, at least the name service
(“NameService”) is available. However, the array might be empty. One
of these strings is passed into the second call, to get back an object
reference to the desired service. The following code segment finds the
reference for the naming service, and narrows it down to the correct
type:
org.omg.CORBA.Object objRef =
orb.resolve_initial_references("NameService");

NamingContext ncRef =
NamingContextHelper.narrow(objRef);

Java Interface Definition Language (JavaIDL) 4-37

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
Bootstrapping the Client Application

Use the COS Naming Service to Retrieve the Remote Object

References

The following code segment retrieves a named object reference from

the naming service, and narrows it to the correct type:
NameComponent nc = new NameComponent("Echo", "");
NameComponent path[] = {nc};
Echo echoRef =
EchoHelper.narrow(ncRef.resolve(path));

A complete name consists of one or more NameComponent elements,

arranged in an array. This whole name is resolved into an object
reference. The actual object name is the last element in the array. The
other elements form the name context. For example, imagine the
elements of the array printed out and separated by a period:
acme.switzerland.zurich.headoffice.floor2.printer21. In this
example, the name is bound to the “root” context: there is only one
element in the array.

The second parameter in the construction of a NameComponent can be

used freely to add a description to the name, such as “executable” or
“object_code”. There is no standard on how to use this property.

4-38 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
Exercise: Bootstrapping/COS Name Server

Exercise objective – Find out which initial services are available in

JavaIDL and (if available) in Netscape Communicator. In addition,
explore the COS naming service.

Preparation
Copy EchoApplet.java and EchoApplet.html to
ServicesApplet.java and ServicesApplet.html.

Tasks

Develop an Applet That Lists the Initial Services Available

Use ServicesApplet.html and ServicesApplet.java as templates

to develop an applet that lists the initial services available to an ORB.

1. Change the directory to labfiles/mod4-jidl/lab/boot.

2. Develop the applet.

3. Try this applet in appletviewer and (if available) in Netscape

Communicator. What services did you find in which environment?
✓ JavaIDL supports NameService, Netscape Communicator so far (v4.51) does not support
any services.

Explore Some Specifics of the COS Naming Service

1. See what happens if you use the second field in naming (kind)

▼ When binding the name (modify EchoServer.java)

▼ When looking up the object (modify EchoClient.java)

▼ When looking up the object, but use a wildcard (“*”) as the kind
✓ Surprisingly, both the name and the kind have to match, and wildcards do not work. This
is surprising because the kind is only a further description of the object, and not really a
part of the name.

Java Interface Definition Language (JavaIDL) 4-39

Explore Some Specifics of the COS Naming Service (Continued)

▼ When binding the object a second time, with the name

differing only in the kind part (modify EchoServer.java;
warning: use “bind”, not “rebind”, as the latter replaces
already existing bindings)
✓ As the kind is only a description of the object, and not really part of the name, a second
bind throws an AlreadyBound exception.

4-40 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
Exercise Summary
Discussion – Take a few minutes to discuss what experiences, issues,
or discoveries you had during the lab exercises.

● Experiences
✓ Ask students what their overall experiences with this exercise have been. You may want to
go over any trouble spots or especially confusing areas at this time.

● Interpretations
✓ Ask students to interpret what they observed during any aspects of this exercise.

● Conclusions
✓ Have students articulate any conclusions they reached as a result of this exercise
experience.

● Applications
✓ Explore with students how they might apply what they learned in this exercise to
situations at their workplace.

Java Interface Definition Language (JavaIDL) 4-41

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
IDL-to-Java Programming Language Mapping Details

With IDL, you can declare items that are unknown to the Java
programming language, such as constants, structs, or variable size
arrays. Because there are several ways to implement these declarations
using constructs known to the Java programming language, a standard
on exactly how the IDL-to-Java programming language mapping is
done is needed. This section covers the details of this standard,
including how each IDL keyword maps to the Java programming
language.

Note – This section is not a tutorial on IDL. To learn IDL, consult a

book covering CORBA.

The following (artificially constructed) IDL contains all the keywords

and constructs that are described on the next pages:
1 module EchoApp {
2 interface Echo {
3 attribute string name;
4 readonly attribute string internalname;
5 string sayEcho(in string myName)
raises (IDLException);
6 };
7
8 exception IDLException {
9 string reason;
10 };
11
12 struct Address {
13 string name;
14 boolean grownUp;
15 };
16
17 typedef sequence<Address, 5> MaxFiveAddresses;
18 typedef sequence<Address> LotsOfAddresses;
19
20 enum SpecialBool {yes, no, maybe};
21 };

4-42 Distributed Programming With Java Technology

Module Construct
The module construct is used to scope IDL identifiers. Here EchoApp is
the enclosing scope:
1 // IDL
1 module EchoApp {
2 ...
3 };

In the Java programming language, a module maps to a package.

1 // generated Java
4 package EchoApp;
5 ...

Java Interface Definition Language (JavaIDL) 4-43

Interface Construct
An interface defines the basic IDL service. Interfaces are a collection of
attributes, exceptions, and operations that can be requested of an object
by a client. JavaIDL maps IDL interfaces to a Java interface. The issue
is that IDL supports multiple inheritance of interfaces only (as IDL
defines interfaces only) while the Java programming language
supports multiple inheritance of interfaces, it does not support
multiple inheritance of classes.

Interface Example
Consider the following example of an IDL interface:
1 // IDL
2 module EchoApp {
3 interface Echo {
4 ...
5 };
6 };

4-44 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
Compiled with idltojava, the following files are generated (on the
client side):

● A Java interface file Echo.java that defines methods that

correspond to the operations in the Echo interface:
// generated Java
package EchoApp;
public interface Echo extends org.omg.CORBA.Object {
...
}

● A Java class file EchoHelper.java that defines methods used to

assist the ORB. The most notable method is narrow, which is used
to cast a CORBA Object type to its appropriate Java object type:
// generated Java
package EchoApp;
public final class EchoHelper {
...
public static EchoApp.Echo
narrow(org.omg.CORBA.Object that)
throws org.omg.CORBA.BAD_PARAM {
...
}
}

● A Java class file EchoHolder.java, which defines methods for

setting and getting the values of an Echo object referred to through
out and inout parameters:
// generated Java
package EchoApp;
public final class EchoHolder
implements org.omg.CORBA.portable.Streamable{
...
}

Java Interface Definition Language (JavaIDL) 4-45

Operations and Parameter Declarations

Operations

Operation declarations in IDL are similar to C function declarations.

An operation consists of the following:

● The return type of the operation or void

● An identifier that names the operation in its scope

● Zero or more parameters

● An optional raises clause

Parameters

An operation parameter can be of any IDL basic or user-defined type.

It must have a directional attribute that informs the communications
service in both the client and server of the direction in which the
parameter is to be passed. Parameters consist of the following:

4-46 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
● in – The parameter is passed from client to server.

● out – The parameter is passed from server to client.

● inout – The parameter is passed in both directions.

The Java programming language does not support passing by value1,

so out and inout do not map directly for primitive types. The Holder
classes are created to support all out and inout operations and return
values.

1. Technically, if you pass an object by value in the Java programming language, you
pass a copy of the reference variable. Therefore it is true that the Java programming
language does not support passing by value.

Java Interface Definition Language (JavaIDL) 4-47

Attribute Declaration
An interface can have attributes as well as operations. IDL attributes
are similar to JavaBeans properties: an attribute definition is logically
equivalent to declaring an accessor and a mutator function. These,
respectively, retrieve and set the value of the attribute. The optional
readonly keyword indicates that there is only an accessor function.

Consider the following IDL interface:

module EchoApp {
interface Echo {
attribute string name;
readonly attribute string internalname;
string sayEcho(in string myName);
};
};

4-48 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
This produces the following Java interface:
package EchoApp;
public interface Echo extends org.omg.CORBA.Object {
String name();
void name(String arg);
String internalname();
String sayEcho(String myName);
}
✓ Note that you will need to implement the actual “name” and “internalname” data field in
the implementation class (servant) and fill in the methods to return the contents of the
attributes.

Java Interface Definition Language (JavaIDL) 4-49

Raises Expressions and Exceptions

Raises Expressions

A raises expression specifies which exceptions can be raised as a

result of an invocation of the operation. The syntax is:
raises ( MyExc1 [, MyExc2 ... ] )

The exceptions raised by an operation can be operation-specific or

those from the set of standard exceptions. The latter might not always
be listed in a raises expression (analogous to a subclass of a Java
programming language RuntimeException, which does not have to
be in the throws clause either).

All the CORBA runtime exceptions are unchecked. Compare this to

RMI: every remote method call in RMI must be enclosed in a try{} /
catch (RemoteException){} brace.

4-50 Distributed Programming With Java Technology

An exception declaration permits the declaration of a struct-like

data structure that might be returned to indicate that an exceptional
condition has occurred. Unlike the Java programming language, an
exception is not a class, but a datatype. The syntax is:
exception <identifier> "{" <member>* "}"
✓ What would happen if an exception were an object? The event object cannot be sent back
with CORBA (no objects by value), only a remote reference to it. To make things worse:
since CORBA does not have a remote garbage collector, and since exception objects tend
to be very short lived, you would soon have a lot of no-longer-used exception objects
clogging up the server. A datatype exception, on the other hand, can be sent back to the
client by value, and can be immediately deleted on the server after sending it off.

An exception has an identifier and zero or more returned member

values. If an exception is returned from a request, the identifier is
accessible; if members are declared, they are accessible. Otherwise, no
further information is available.
module EchoApp {
...
exception IDLException {
string reason;
};
};

JavaIDL exceptions extend the omg.org.CORBA.UserException class:

1 package EchoApp;
2 public final class IDLException
3 extends org.omg.CORBA.UserException {
4 public String reason;
5 public IDLException() {
6 super();
7 }
8 public IDLException(String __reason) {
9 super();
10 reason = __reason;
11 }
12 }

Java Interface Definition Language (JavaIDL) 4-51

Thetypedef Keyword
IDL provides constructs for naming data types. The typedef keyword
is used to associate a name with a data type.
typedef long IDNumber;
typedef string SSNumber;

The Java programming language has no construct equivalent to the

IDL typedef statement. IDL typedefs for simple types do not map
directly to the Java programming language. Instead, the original type
is substituted for the new type everywhere the new type is
encountered by the idltojava compiler.

4-52 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
Basic Java Technology Types
IDL types map to Java types as shown in Table 4-1.

Table 4-1 IDL Types to Java Types Mappings

IDL Type Java Type

float float
double double
long, unsigned longa int
long long, unsigned long long* long
short, unsigned short* short
unsigned long int
unsigned short int
char, wchar char
boolean boolean
octet byte
string, wstring java.lang.String
enum, struct, union class
a. It is up to the developer to maintain the accuracy of unsigned numbers
because Java technology does not support unsigned types.

Java Interface Definition Language (JavaIDL) 4-53

The struct Keyword

The IDL struct keyword is used to contain a collection of data that
you would like to pass as a single item. The struct keyword is
mapped to a class that provides instance variables for the fields and a
constructor that takes a value for each variable.
//IDL
module EchoApp {
...
struct Address {
string name;
boolean grownUp;
};
};

// generated Java
package EchoApp;
public final class Address {
//instance variables
public String name;
public boolean grownUp;

4-54 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
//constructors
public Address() { }
public Address(String __name, boolean __grownUp) {
name = __name;
grownUp = __grownUp;
}
}
✓ You have already heard several times that it is not possible to pass objects over the wire,
only references; and now structs are mapped to objects. How can this Java object travel
from a server to a client and vice versa? The Java object does not travel. What goes over
the wire is a CORBA struct, not a Java object. If JavaIDL “decides” to map the incoming
CORBA struct to a local Java object, this is fine. All it has to do later, once the CORBA
struct travels back, is to map the Java object to an outgoing CORBA struct again.

Java Interface Definition Language (JavaIDL) 4-55

The Sequence Keyword

A sequence is like an array, with two key differences:

● It is one-dimensional.

● It does not have a fixed size. Optionally, it can have a maximum

size (which is set at compile time).

A sequence can be unbounded or bounded. The bounded sequence is

the one with a maximum size.
typedef sequence<Address, 5> MaxFiveAddresses;
typedef sequence<Address> LotsOfAddresses;

4-56 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
Sequence Mapping
Sequences are mapped to Java programming language arrays (“Java
arrays”). The idltojava compiler generated a helper and a holder
class for each sequence, but these classes are used only internally by
the stub and skeleton. For the developer, IDL sequences look like
regular Java arrays.

However, there is a distinct difference between a sequence and an

array: An array does have a fixed size, which is set when the array is
constructed. The sequence, on the other hand, does have a length,
which is determined at runtime.

This difference is not a problem when a sequence is retrieved after a

call. The sequence is unmarshalled into an array that has exactly the
right size. When passing a sequence, you must pass an array that has
the correct size. If the size cannot be determined at compile time, you
can use Java programming language vectors, which can grow
dynamically. At the moment of passing the sequence, the vector must
be converted into a correctly sized array.

Java Interface Definition Language (JavaIDL) 4-57

Array
IDL arrays are mapped in the Java programming language in the same
way as the bounded sequence. The difference is that while a bounded
sequence can accept arrays smaller than its bound, the array must be
exactly the same size. The size is checked when the array is marshalled
as an argument to an IDL operation.

4-58 Distributed Programming With Java Technology

The enum Construct

The IDL enum constructs are mapped to Java classes with a single
static final variable for each member of the enumerated type.
//IDL
enum SpecialBool {yes, no, maybe};

The following is the Java code generated from the IDL:

package EchoApp;
public final class SpecialBool {
public static final int _yes = 0,
_no = 1,
_maybe = 2;
public static final SpecialBool yes = new
SpecialBool(_yes);
public static final SpecialBool no = new
SpecialBool(_no);
public static final SpecialBool maybe = new
SpecialBool(_maybe);
public int value() {
return _value;
}

Java Interface Definition Language (JavaIDL) 4-59

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
public static final SpecialBool from_int(int i)
throws org.omg.CORBA.BAD_PARAM {
switch (i) {
case _yes:
return yes;
case _no:
return no;
case _maybe:
return maybe;
default:
throw new org.omg.CORBA.BAD_PARAM();
}
}
private SpecialBool(int _value){
this._value = _value;
}
private int _value;
}

4-60 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
Exercise: IDL-to-Java Programming Language Mapping Details

Exercise objective – Explore details of the IDL to Java programming

language mapping; specifically, the difficulties with the IDL sequence
to Java programming language array mapping.

Tasks

Convert a JDBC ResultSet to an IDL Sequence

A JDBC ResultSet is a datastructure with an unknown length (a

typical Java programming language Enumeration). This matches
perfectly to an IDL sequence, which does not have a fixed length
either. However, the IDL sequence is mapped to a Java programming
language array, not to some variable length Java programming
language datastructure. Your job is to convert the JDBC ResultSet
into a correctly sized array of Coffee (that is, the array must not contain
null values).

1. Change the directory to labfiles/mod4-jidl/lab/jdbc.

2. Look at the IDL file DBLookup.IDL. It describes a server object that

is capable of doing a database lookup for you and sending back
the entire result set as a “sequence of coffee.” The datastructure
“coffee” is also declared in DBLookup.IDL.

The IDL, the server, and the client are already written and do not
need to be changed.

3. Compile the servant and run it. It does a DB lookup, but does not
use the retrieved ResultSet to fill in the sequence. Compile and
run the provided application to confirm that it is working.

4. Replace the existing code with code that converts the ResultSet
into an array of Coffee! (Hint – Use a Java programming language
Vector temporarily.)

5. Start the tnameserv, server, and client.

✓ When students get the database connection error on the server console, remind them of
the JDBC exercises and have them change the host in CoffeeHostImpl.java.

Java Interface Definition Language (JavaIDL) 4-61

Explore Specifics of the IDL-to-Java Programming Language

Mapping (Optional)

In this exercise, you use an IDL containing the most important IDL
keywords: MappingTest.IDL. This IDL can be used to test the whole
range of the IDL to Java programming language mapping
specification.

1. Change the directory to labfiles/mod4-jidl/lab/idl.

2. Compile the IDL file, and look at the various generated files,
especially MInterface.java. MStruct.java,
MSequenceHolder.java, MEnum.java and MEnumHolder.java,
because these are the classes that are used in the source code.

3. Compile the application provided and start it. There is an error

with the attribute. Obviously an attribute declaration in IDL alone
does not suffice.

4. Extend the servant to support the declared attribute

(MAttribute).

5. Change the client so that the custom exception gets fired (look at
the servant code to see how to trigger this). Even if you fill in the
arguments on the servant before throwing the exception, they are
not sent back to the client.

6. Change the client back to the previous state.

7. Try to break the servant in several ways:

▼ Instead of using the Holder you get, create a new one (that is,
replace theMSequenceHolder.value = new ... with
theMSequenceHolder = new MSequenceHolder...). What
happens?
✓ A CORBA internal error is thrown. You are allowed to replace the value of a holder, but
never the holder itself.

▼ The sequence is bounded, as you can see in the IDL file. Try to
give back too many elements. What happens?
✓ A CORBA marshalling error, org.omg.CORBA.MARSHAL, is thrown.

4-62 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
▼ Give back an incorrectly sized array. Use an array of size 2, but
fill in only the first element; the second one is initialized with
null. What happens?
✓ CORBA internal error, org.omg.CORBA.UNKNOWN, gets thrown. A sequence, by definition,
cannot contain null elements. This is analogous to a Java vector. Unfortunately, a Java
array can contain null elements, so you have to pay close attention. Why the sequence got
mapped to a Java array and not to a vector, is not known.

▼ Try to overwrite the in argument in the servant. Does it work?

Does the argument change on the client too?

Java Interface Definition Language (JavaIDL) 4-63

● Experiences
✓ Ask students what their overall experiences with this exercise have been. You may want to
go over any trouble spots or especially confusing areas at this time.

● Interpretations
✓ Ask students to interpret what they observed during any aspects of this exercise.

● Conclusions
✓ Have students articulate any conclusions they reached as a result of this exercise
experience.

● Applications
✓ Explore with students how they might apply what they learned in this exercise to
situations at their workplace.

4-64 Distributed Programming With Java Technology

Objects as Parameters in Remote Calls

So far, only primitive datatypes have been used as arguments in the

remote calls. These arguments were passed by value to the server
implementation, inside a holder class if necessary. In this section you
will pass an object as an argument. Use the same setup as was used in
Module 3, “Remote Method Invocation (RMI),” for the Agent example.
To start, the following are the IDLs:
module AgentApp {
interface Agent {
void run();
long getResult();
};
};

module WorkerApp {
interface Worker {
void accept (inout AgentApp::Agent agent);
};
};

Java Interface Definition Language (JavaIDL) 4-65

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
Two remote objects are declared: the Worker is the servant, and the
Agent is the class that is passed as an argument. This is analogous to
the two Java interface declarations, Worker.java and Agent.java, in
Module 3.

The source code is derived from the RMI example, so it looks similar.
One difference is in the IDL. It contains an additional method, long
getResult. You will see later why you need this method. Another
difference is the requirement for a holder class: inout arguments get
mapped this way. The following is a section from
WorkerClient.java:
Agent agent = new CalcFactorial(10);
AgentHolder holder = new AgentHolder(agent);
workerRef.accept(holder);
System.out.println("Returned from remote call\n");
agent = holder.value;
System.out.println("Result: " + agent.getResult());

Unexpectedly, this code seems to work. The output of the application

is the same as with RMI. Did the Agent travel over the wire after all?
That is, was it passed by value to the server? Find the answer yourself
by doing the following exercise.

4-66 Distributed Programming With Java Technology

Exercise objective – Confirm that objects by value do not work with

JavaIDL by trying to duplicate the Agent example from Module 3 with
JavaIDL.

Tasks

Solve the Agent Example From RMI With JavaIDL and Find the
Differences From RMI

Complete the following steps:

1. Change the directory to labfiles/mod4-jidl/lab/agent.

2. Compile and run the application provided. Look at the source

code and compare it to the RMI agent example. Locate the one
crucial difference.
✓ In JavaIDL, the Agent is declared as a remote object; there is an IDL for it. idltojava was
run against the IDL, and the actual implementation (CalcFactorial) extends
_AgentImplBase. This means that whenever you have an Agent as a parameter in a remote
call, a remote reference to the Agent is passed. In RMI, on the other hand, the Agent is not
a remote object. The interface Agent.java does not extend Remote, and rmic was not run
against CalcFactorial.java (which implements the Agent). This means that whenever
there is an Agent as a parameter in a remote call (such as a CalcFactorial), a copy of the
Agent is passed.

3. Find the consequences of the difference. On which machine does

the Agent execute? If you think you know the answer, prove it by
modifying the source code.
✓ The Agent always runs on the (virtual) machine it was created on, since it cannot leave it.
In step 2, it runs on the client, which is exactly what was not wanted.

4. Modify the servant (WorkerImpl.java) to do the following:

a. Create a new CalcFactorial.

b. Calculate the result (call its run method).

c. Assign the new object to the holder-class (Agent.value =

... in the source code).

Java Interface Definition Language (JavaIDL) 4-67

To test your knowledge of ‘object by reference,’ ask yourself the

following questions:

● On which machine does the CalcFactorial run this time?

✓ As before, the CalcFactorial runs on the (virtual) machine it was created on, only this
time, this machine is the server.

● On what class does the holder “hold” the client before returning:
CalcFactorial.class or _AgentStub.class?
✓ CalcFactorial.class. After all, you just created this object and assigned it to the holder’s
value.

● On what class does the holder “hold” the client after returning:
CalcFactorial.class or _AgentStub.class?
✓ _AgentStub.class. A remote reference to the newly created CalcFactorial on the server is
passed back.

4-68 Distributed Programming With Java Technology

● Experiences
✓ Ask students what their overall experiences with this exercise have been. You may want to
go over any trouble spots or especially confusing areas at this time.

● Interpretations
✓ Ask students to interpret what they observed during any aspects of this exercise.

● Conclusions
✓ Have students articulate any conclusions they reached as a result of this exercise
experience.

● Applications
✓ Explore with students how they might apply what they learned in this exercise to
situations at their workplace.

Java Interface Definition Language (JavaIDL) 4-69

The preceding exercise is based on the following principle: The

WorkerServant gets an AgentHolder, which contains an Agent.
Remember, Agent is just an interface. What class does the
WorkerServant really get? The answer is _AgentStub. Obviously, the
AgentServant does not get a copy of the Agent, but a remote
reference to the Agent residing in the client. This is the crucial
difference from RMI. In the RMI example, the WorkerServant gets a
copy of the object; that is, a CalcFactorial.

If you call run from the WorkerServant code, you will do a remote
call, this time back to the client. This is what is known as a callback. The
server calls back to the client, and the roles are reversed.

CORBA does not allow passing of objects by value. Whenever there is

an object as a parameter in a method call, a remote reference is passed.

Now that you know this rule, what happens if you create a new
CalcFactorial on the server, and pass this object back inside of the
holder class? Again, the rule holds. You cannot send back a copy of the
object to the client; it has to be a remote reference. An unusual situation
now exists. You had a local object on the client before the accept call,
and a remote reference right after the call. This gets confusing, for both
you as the developer and for CORBA. The ORB on the server keeps a
reference to the newly created CalcFactorial, because CalcFactorial is a
CORBA object. Theoretically, it could be dropped once the server ORB
“knows” that no remote references to this object exist anymore (which
is the case as soon as the client exits). This behavior is known as remote
garbage collection. However, JavaIDL does not implement remote
garbage collection, and so the object stays on the server forever,
wasting resources.

This callback is also the reason why you needed the additional method
getResult. The server created a CalcFactorial, but the client does
not “know” this. As far as the client is concerned, it gets back a remote
reference to an Agent, not to a CalcFactorial. (You do not even
have an IDL for a CalcFactorial.) If the Agent does not have a
method getResult, the client cannot get at the result of the
calculation. Compare this to the RMI case. There, you got back a copy
of an Agent, which happened to be a CalcFactorial. Because you do
have the copy back on the client, you can cast it to any class this Agent
happens to be.

4-70 Distributed Programming With Java Technology

Futures

Objects by Value

You have seen several times that you cannot transmit CORBA objects
by value. This is going to change; the next version of CORBA, 3.0, is
supposed to contain this enhancement. A draft specification is already
under review. You can find the FTP address of the draft document in
the “Additional Resources” section of this module.
✓ At http://www.omg.org/news/pr98/compnent.html you can also find more current
information on the most current version of CORBA.

RMI Over IIOP

Sun Microsystems is working with IBM on an implementation of RMI

that uses the CORBA wire protocol (IIOP) instead of its own. This will
allow interoperability of CORBA and RMI: an RMI server can be
called from a CORBA client, and a CORBA server can be called from
an RMI client. To support this as transparently as possible, IIOP must
allow objects by value. This is the main reason why the objects by value
proposal has been made.

Java Interface Definition Language (JavaIDL) 4-71

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
4
The “Additional Resources” section at the beginning of this module
cites a JavaOne session about RMI over IIOP.

Portable Object Adapter/Object Activation

So far, JavaIDL allows you to write only simple, transient servers.

Object activation, as you have seen with RMI, is not possible with the
current JavaIDL beta. Other commercial ORB products allow more
feature-rich servers to be written. However, because there was no
sufficiently exact standard on how to do this prior to CORBA 2.2,
servers are not portable between different ORB products.

CORBA Version 2.2 includes the definition of a portable object adapter

(POA). Using this adapter, you can write feature-rich servers in an
ORB-independent fashion. If and when JavaIDL will include the POA
is not known at this time.

Another possibility for writing better servers is to use RMI. Once the
RMI over IIOP extension is available, you can write servers using RMI,
with little CORBA knowledge. Because RMI supports object activation
today, you would not have to wait for POA to arrive in JavaIDL.

4-72 Distributed Programming With Java Technology

Check that you were able to accomplish or answer the following

questions related to content in this module:

❑ Describe the basic CORBA object management architecture

❑ Describe the role of JavaIDL in relation to other commercial Java

CORBA products

❑ Create and deploy a JavaIDL server object and a JavaIDL client

application

❑ Describe how the JavaIDL bootstrapping process works

❑ Describe how IDL is mapped to the Java programming language

❑ Explain why the RMI Agent example does not work with the
current CORBA

Java Interface Definition Language (JavaIDL) 4-73

How could you use JavaIDL to distribute your applications?

4-74 Distributed Programming With Java Technology

Objectives
Upon completion of this module, you should be able to:

● Understand the basic concepts of servlets

● Create your own servlets for Web-based applications

● Compare HTTP servlets to other technologies, such as common

gateway interface (CGI)

Servlets are used to extend or implement server functionality. The

Servlets API enables developers to implement servlets that can be used
by any server enabled for the Java programming language.

Discussion – Consider the following questions:

● What are the advantages of server-side components over classic

mainframe servers?

✓ Servers become easier to understand, to maintain, and to extend when written in a

modular way using components.

● What makes the Java programming language an ideal platform for

server applications?
✓ Platform neutrality, integrated thread model, garbage collection, and its security model.

5-2 Distributed Programming With Java Technology

Additional resources – The following resources can provide

additional detail on the topics presented in this module:

● Java Server Product. [Online]. Available:

http://java.sun.com/products/java-server

● Java Servlets. [Online]. Available:

http://java.sun.com/products/java-server/servlets

● Java Web Server. [Online]. Available:

http://java.sun.com/products/webserver/index.html

Servlets Overview

Servlets are protocol- and platform-independent server-side

components written in the Java programming language. Servlets are
used in servers enabled for the Java programming language to extend
the services the server provides. Because you can load servlets
dynamically, they extend a server’s functionality at runtime.

Because servlets run inside a server, they are the server-side

counterpart to applets. Servlets are Java application components that
you can download on demand to the server that needs them.

Servlets were first developed for the Java Web Server™. The Servlet
API is generic enough to implement any request-response type of
server, although today’s main usage are Web-based services.

5-4 Distributed Programming With Java Technology

The Servlets API is a standard Java programming language extension

and is bundled with the Java 2 SDK in the javax.servlet package.
The package core consists of six interfaces, two stream classes, and two
exception classes. A servlet must implement the Servlet interface. For
easy and fast development, an implementation of the Servlet
interface is provided within the package.

The server calls the servlet’s service method to handle a request from
a client. Servlets are running in a multithreaded environment and
many requests to a servlet’s service method can be made
simultaneously. Therefore, servlets must be implemented in a thread-
safe way. To simplify thread-safe servlets development, a servlet can
implement the SingleThreadModel interface. A single-threaded
servlet’s service method is never called more then once at a time. The
server ensures that the service is never called several times
simultaneously.

Figure 5-1 on page 5-6 shows the class hierarchy for the
javax.servlet package.

The javax.servlet Package

java.lang.Object

Servlet
GenericServlet

ServletConfig

ServletContext

ServletRequest

ServletResponse

SingleThreadModel

java.lang.Exception
java.lang.Throwable
ServletException

UnavailableException
Legend
java.io.InputStream Class
ServletInputStream Abstract
Class

java.io.OutputStream Interface
Extends
ServletOutputStream
Implements

Figure 5-1 The javax.servlet Package

5-6 Distributed Programming With Java Technology

Simple Servlet
The implementation of a simple servlet is outlined in the following
example. It extends the GenericServlet class included in the
Servlet package. The GenericServlet implements the Servlet
interface and eases development of servlets, because an
implementation of every method defined by the interface is provided.
1 import java.io.IOException;
2 import javax.servlet.*;
3
4 public class SimpleServlet extends GenericServlet {
5 public void service(
6 ServletRequest request,
7 ServletResponse response)
8 throws ServletException, IOException
9 {
10 // The Servlet's functionality is
11 // implemented here.
12 }
13 }

Servlet Interaction
The service method is provided with two parameters, Request and
Response, which encapsulate the interaction with clients or chained
servlets. The request parameter contains all data sent by the client,
including status and meta information. The response parameter is
used to send a result back to the client. Usually the data is read from
an InputStream, processed and written to an OutputStream. The
SimpleServlet’s service method retrieves the two streams as follows:
ServletInputStream in =request.getInputStream();
ServletOutputStream out =response.getOutputStream();

The data format used to communicate is not specified within the API.
It is up to the user to specify how requests and responses are encoded.
The javax.servlet.http package defines new Request and
Response interfaces, which are used to implement HTTP
communication channels.

5-8 Distributed Programming With Java Technology

HTTP Servlets

HTTP servlets are used by Web servers enabled for the Java
programming language to build interactive Web applications, generate
dynamic Web pages, or integrate databases or other systems into the
Web. The servlets are loaded upon request into the running Web
server in response to client requests.

HTTP servlets are Java technology components (“Java components”)

and therefore do not have memory leaks or illegal memory accesses
that might occur with components written in C or C++. The HTTP
servlet is loaded once into the server and can afterwards serve many
requests, even simultaneously. This approach differs much from the
Common Gateway Interface (CGI), where for each request a new
process needs to be started, leading to delays and placing an excessive
burden on the Web server.

The Java Web Server provided by Sun was the first Web server that
used servlets to build complex interactive Web applications. Today,
many other Web servers are enabled for the Java programming
language servlets.

Servlets 5-9
Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
5
Note – You can extend Netscape, Microsoft, and other web server
products with servlet support.

The javax.servlet.http Package

The javax.servlet.http package is a standard Java programming
language extension and is bundled with the javax.servlet package.
It contains interfaces and classes to handle HTTP- and HTML-specific
communication. Furthermore the package offers HTTP session
management and tracking using either cookies or automatically
generated URLs. The HTTPServletRequest interface offers access to
HTTP specific header and protocol information.

This section examines the HttpServlet class, which offers developers

a good starting point to build servlets for Web servers.

5-10 Distributed Programming With Java Technology

Servlet
GenericServlet

HttpServlet

HttpSession

EventListener
HttpSessionBindingListener

HttpSessionContext

ServletRequest

HttpServletRequest

ServletResponse
Legend
HttpServletResponse Class
Abstract
java.util.EventObject Class
Interface
HttpSessionBindingEvent
Extends
HttpUtils Implements

Figure 5-2 The javax.servlet.http Package

Servlets 5-11
Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
5
HTTP Servlet Example
The following example servlet prints the date and time formatted in
HTML to its response stream. The result displayed by a Web browser
looks similar to Figure 5-3.
1 import java.io.*;
2 import java.util.Date;
3 import javax.servlet.*;
4 import javax.servlet.http.*;
5
6 public class DateServlet extends HttpServlet
7 {
8 public void doGet(HttpServletRequest request,
9 HttpServletResponse response)
10 throws ServletException, IOException {
11
12 // set response header fields first
13 response.setContentType("text/html");
14
15 // then write the data of the response
16 PrintWriter out = response.getWriter();
17
18 out.println("<HTML><HEAD><TITLE>");
19 out.println(getServletInfo());
20 out.println("</TITLE></HEAD><BODY>");
21 out.println("<H1>"+getServletInfo()+"</H1>");
22 out.println("<P>This Page has last been " +
23 "accessed on "+new Date()+".");
24 out.println("</BODY></HTML>");
25 out.close();
26 }
27
28 public String getServletInfo() {
29 return "Date Servlet";
30 }
31 }
32

5-12 Distributed Programming With Java Technology

Figure 5-3 DateServlet’s output

Exercise objective – Running your first servlet called DateServlet.

✓ As of JSDK2.1, servletrunner has been superseded with the startserver command

which contains a shell script for launching the ‘runner’.

Preparation
1. Change the directory to labfiles/mod5-servlets/lab and
compile DateServlet.java.

2. Copy DateServlet.class to the Java Servlet Development Kit’s

/servlets directory, /jsdk2.1/webpages/WEB-INF/servlets.

3. Copy DateServlet.html to the /webpages directory.

4. Set the server name and port in /jsdk2.1/default.cfg, and

execute the startserver command.
✓ More explicit instructions and troubleshooting information can be found in
SL301_SOL_LF/labfiles/mod5-servlets/lab/README.

Tasks
Run DateServlet and test its functionality.

1. Open the /jsdk2.1/webpages/DateServlet.html page in

Netscape and click on the link to the DateServlet.

2. Click on the Reload button several times and watch for the
different response times, from the first to subsequent requests.
Also check the date and time change for each request. Observe
the different response times by shutting the server down and
restarting it. Click on Reload in the browser again.

5-14 Distributed Programming With Java Technology

Run DateCounterServlet

Servlets can save states between several requests. This is shown by

adding a counter to the servlet and printing it on the Web page. You
see how many times the page (the servlet) has been accessed.

Note – The servlet counts every request and not just requests from one
person or machine. The exercise in this module shows how accesses
from individual persons are counted.

1. Copy DateServlet.java to DateCounterServlet.java.

2. In the new file, change the class name from DateServlet to

DateCounterServlet.

3. Add a counter as an instance variable to the servlet.

4. Add the counter’s value to the HTML output.

5. Compile DateCounterServlet and test it using your browser

and DateCounterServlet.html. (Note – You do not have to
restart the server. The newly generated class file is loaded on
demand.)

Servlets 5-15
Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
5
Exercise Summary
Discussion – Take a few minutes to discuss what experiences, issues,
or discoveries you had during the lab exercises.

● Experiences
✓ Ask students what their overall experiences with this exercise have been. You may want to
go over any trouble spots or especially confusing areas at this time.

● Interpretations
✓ Ask students to interpret what they observed during any aspects of this exercise.

● Conclusions
✓ Have students articulate any conclusions they reached as a result of this exercise
experience.

● Applications
✓ Explore with students how they might apply what they learned in this exercise to
situations at their workplace.

5-16 Distributed Programming With Java Technology

Using an HTTP Servlet

HTTP servlets are usually invoked using a Web browser. Servlets only
providing data, such as the DateServlet shown before, are referenced
by a URL, such as http://localhost:8080/servlet/dateservlet.
The exact path is determined by the administrator of the server and by
the server’s architecture. The Java Web Server, for example, uses the
path /servlet/servletname to reference servlet servletname.
The user either enters the URL or clicks on a hypertext link that leads
to the servlet.

HTTP servlets that require user input are embedded in an HTML

form. The servlet to be used to process the data entered is referenced
by a usual HTML form tag. The class
javax.servlet.http.HttpServlet supports only the post method to
receive the data entered in the form.

Example Usage
When dynamic HTML pages must be created “on the fly” (while the
program is running) or HTML forms must be processed, you can use
HTTP servlets. For HTML page generation, two approaches are used:

● As in the DateServlet example, a complete page is generated by

the servlet. The disadvantage is that any change of the static part
of the HTML page must be accomplished by editing and
recompiling the servlet.

● Some Web servers offer Server Side Includes (SSI), where the Web
server parses the requested HTML pages. Special HTML tags can
denote servlets, which are executed by the server. The HTML tag
is replaced by the output of the servlet.

Many Web applications use some sort of HTML form processing, such
as feedback or registration pages, online shopping, or access to other
systems.

5-18 Distributed Programming With Java Technology

Servlet Life Cycle

Servlets have their own life cycle that is independent of the server’s
life cycle. You can load and unload servlets at runtime. On the first
request for a specific servlet, the server loads the appropriate class files
either from a local disk or through the network. Exactly one instance
of the servlet is created and the servlet’s init method is called. A
ServletConfig object is passed along to the servlet. If the servlet does
not throw an UnavailableException, it is ready to answer requests
through the service method. The server ensures that no service
method call is performed before init returns.

Servlets can be unloaded at runtime. Unloading means unreferencing

and garbage collecting the servlet and also removing the servlet’s class
file from the Java virtual machine. This enables servers to reload, for
example, a newer version of the same servlet without restarting the
server. Before a servlet is unloaded, the server calls the destroy
method. Destroy removes all resources allocated either during
runtime or by init. Destroy can be called before all service requests
are completed. The servlet must either wait for any pending requests to
complete or abort the requests. After destroy returns, the servlet is
unloaded.

HTTP Servlet Request

The HttpServletRequest interface extends the underlying

ServletRequests class by adding methods to access HTTP-specific
header and status information, such as current session or
authentication type information. This information is provided with
every HTTP request.

To read data entered in an HTML form, two different approaches are

offered:

● Obtain a stream or a reader object from which the data sent by the
Web browser can be read. The input gathered this way must be
parsed according to the HTML specification to get the data entered
into the form.

5-20 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
5
● Use the various getParameter methods to get to the form data
directly. This way the HTTP servlet parses the data sent by the
Web browser and builds a dictionary using the field names as key
and the form entries as values.

The two approaches cannot be mixed. Using the getParameter

method is usually the simpler way of processing a request as the
parsing is done by the request object.

Exercise objective – Access additional HTTP header fields in a

request.

Tasks

Access the Snoop Servlet and Study Its Output

Complete the following steps:

1. Change the directory to labfiles/mod5-servlets/lab.

2. Compile SnoopServlet.java, and copy the resulting .class file

to /jsdk2.1/webpages/WEB-INF/servlets.

3. Modify SnoopServlet.html as needed (modify the URL string),

copy it to /jsdk2.1/webpages, and open it in Netscape. Click
the link to the Snoop servlet and let the servlet snoop your
request.

Run DateLinkServlet

DateCounterServlet gets another feature. Add a link in the HTML

output that points back to the servlet itself like in the Snoop servlet.

1. Copy DateCounterServlet.java to DateLinkServlet.java

and rename the class appropriately.

2. Look at SnoopServlet.java and add a link from the

DateLinkServlet to itself in a similar way.

Why should you not hardcode the link? (Optional)

5-22 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
5
Exercise Summary
Discussion – Take a few minutes to discuss what experiences, issues,
or discoveries you had during the lab exercises.

● Experiences
✓ Ask students what their overall experiences with this exercise have been. You may want to
go over any trouble spots or especially confusing areas at this time.

● Interpretations
✓ Ask students to interpret what they observed during any aspects of this exercise.

● Conclusions
✓ Have students articulate any conclusions they reached as a result of this exercise
experience.

● Applications
✓ Explore with students how they might apply what they learned in this exercise to
situations at their workplace.

Client Interaction

HTTP Servlet Response

The HttpServletResponse interface provides methods to format and
send HTTP-specific header and error codes. The HTTP error codes are
provided as constants in the interface.

To send data to the Web client, a stream or a writer object is provided

by the response object. Before retrieving a stream or a writer from the
response object, you must set the content’s Multipurpose Internet Mail
Extension (MIME) type using the setContentType method.
✓ MIME means Multipurpose Internet Mail Extension. MIME is used for providing a mapping
of what helper application is to be launched when a particular data format (GIF, audio, and
so on) is received.

5-24 Distributed Programming With Java Technology

HTTP Session Management

HTTP requests are stateless. However, there are several methods of
building a session management layer on top of HTTP. You cannot
build applications, such as online shopping without sessions because
you must track user state information throughout multiple page
requests.
✓ For example, when the abstraction of a “shopping cart” is to be provided.

Sessions are either implemented using cookies or special URLs that are
generated on the fly. The session management facilities within the
HTTP servlets package do both in a transparent manner.
Implementations of the Session interface manage sessions. They
provide methods to identify sessions and store and retrieve data
associated with the session.

Session management in conjunction with user authentication and

secured communication (HTTPS) allow you to develop sophisticated
Web-based applications.

Exercise objective – Learn about sessions.

Preparation
This exercise requires a Web browser that supports cookies.

Tasks

Use the Session Servlet

Access the session servlet and compare it to DateCounterServlet.

1. Change the directory to labfiles/mod5-servlets/lab.

2. Copy SessionServlet.html to /jsdk2.1/webpages, open it in

Netscape, and follow the link to the servlet.

3. Try to connect to your neighbor’s server and let your neighbors

connect to yours (if the setup in your classroom allows it). If you
do the same thing with DateCounterServlet, what will be the
difference? Check your answer.
✓ The DateCounterServlet sums up all connections. The session Servlet counts all
accesses per user.

4. Study the source code and find out how session management is
done using HttpServlets.

5-26 Distributed Programming With Java Technology

● Experiences
✓ Ask students what their overall experiences with this exercise have been. You may want to
go over any trouble spots or especially confusing areas at this time.

● Interpretations
✓ Ask students to interpret what they observed during any aspects of this exercise.

● Conclusions
✓ Have students articulate any conclusions they reached as a result of this exercise
experience.

● Applications
✓ Explore with students how they might apply what they learned in this exercise to
situations at their workplace.

Application Designs

The servlets API is generic; therefore, you can build many different
applications with completely different designs. This section looks at
application designs using HTTP servlets.

An application consists of a client, which does user interaction; a

server handling data; and some business or application logic. The Web
browser and the servlet are considered the first tier, because the
browser renders only HTML and feeds data entered by the user, and is
not part of the system to be built.

5-28 Distributed Programming With Java Technology

Two-Tier Design
In a two-tier architecture, the servlet interacts with the users and gets
and sends data to another system: a database, for example. This design
is best suited for providing a Web front end to legacy systems, which
already contain all the business and application logic.

Here servlets correspond directly to applets, which would be a good

alternative, but might be considered inappropriate if a browser
capable of running Java programs is not available on the client side, or
when the applet security model is considered too restrictive.

Of course, you can enhance servlets with business logic and directly
access databases using JDBC. For small applications, with the servlet
as the only client of the database, this design might be a viable
solution. For larger applications with many clients, it would be a bad
choice, because the business logic needs to be rebuilt in the other
clients as well.

Three-Tier Design
Within a three-tier design using HTTP servlets, the business or
application logic is isolated in a middle tier. The middle tier offers a
more abstract API to the top tier using RMI or CORBA, for example.
The top tier provides for user interaction and forwards the user’s
requests to the middle tier by API calls. The top tiers are small and
lightweight and can be implemented in many different ways, such as
full Java programs, applets, programs built using some other
language, and HTTP servlets. The bottom tier is used to store and
retrieve data. Instead of rebuilding legacy systems, a middle tier can
integrate several systems and databases and build new applications on
top of them. The new API is then used by the clients to interact with
the end users.

5-30 Distributed Programming With Java Technology

Exercise objective – Show how data entered in a form could be

processed by an application.

Preparation
Check that the path specified in the servlet.properties for the
LeShopServlet exists. Depending on the setup in your classroom,
your path might be different. Look in the properties file or ask your
instructor.

Tasks

Use the LeShop Servlet to Store Data in a File

Fill out the HTML form in LeShopServlet.html and let the servlet
process the data entered.

1. Change the directory to labfiles/mod5-servletslab.

2. Open LeShopServlet.html in your browser.

3. Fill out the form and send it to the servlet for further processing.

4. Watch the file where the data entered is stored. See how it grows
when more data is entered.

5. Study the source code and learn how data entered in a form is
extracted and processed.

Servlets 5-31
Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
5
Exercise Summary
Discussion – Take a few minutes to discuss what experiences, issues,
or discoveries you had during the lab exercises.

● Experiences
✓ Ask students what their overall experiences with this exercise have been. You may want to
go over any trouble spots or especially confusing areas at this time.

● Interpretations
✓ Ask students to interpret what they observed during any aspects of this exercise.

● Conclusions
✓ Have students articulate any conclusions they reached as a result of this exercise
experience.

● Applications
✓ Explore with students how they might apply what they learned in this exercise to
situations at their workplace.

5-32 Distributed Programming With Java Technology

Before continuing on to the next module, check that you were able to
accomplish the following in this module:

❑ Understand the basic concepts of servlets

❑ Create your own servlets for Web-based applications

❑ Compare HTTP servlets to other technologies, such as CGI

How could you take advantage of servlets in an existing application?

5-34 Distributed Programming With Java Technology

Objectives
Upon completion of this module, you should be able to:

● Understand the fundamentals of the Object Bus model

● Understand its main application areas

● Explain the differences between iBus, RMI, and CORBA

Object Bus systems enable developers to write applications that

interact in a one-to-many and many-to-many fashion by transmitting
events through ubiquitous communication channels.

Discussion – Consider the following questions:

● How do you design an RMI application that distributes near real-

time events, such as stock quotes to a dynamically changing set of
listener applications?

✓ You need to provide a network-centric broker daemon to which the listeners of the events
connect. A listener typically is a graphical user interface on the workstation of a stock
trader. When the talker application has a new stock quote to deliver, it requests that the
broker daemon distribute that event to any registered listener application. Another
solution consists of the receivers setting up an RMI connection to the talker application,
to periodically poll the talker for new quotes.

● What impact does this have on network load?

✓ Each quote results in (at least) one RMI request from the talker to the daemon and in one
RMI per registered listener, to “broadcast” the quote. The polling solution also results in
considerable network traffic.

6-2 Distributed Programming With Java Technology

The Object Bus Model

The basic communication paradigm of CORBA and RMI is “request

and reply.” An application explicitly requests specific data from a
server object by invoking a method on a local communication stub.
However, applications, such as financial data delivery require an
approach where listener applications subscribe for events, and talker
applications push events to their listeners through ubiquitous
communication channels. This is called the Object Bus model.

In this model, an event can cause any kind of (Java technology) object,
(such as a stock quote, an audio data packet, an alert indicating that an
aircraft engine is getting too hot, or even a user interface component)
to be displayed by the listeners.

Object Bus middleware functions much like radio transmission. A

radio station transmits on a certain channel and radio listeners tune
into that channel. The radio station does not need to know its receivers
to transmit. The listeners do not need to know the location of their
transmitter.

Object Bus Systems 6-3

Object Bus Example

Consider the stock quote example. With Object Bus middleware, you
can implement the application without providing a network-centric
broker daemon, and without polling the talker for new events. The
Object Bus provides ubiquitous communication channels to which
listeners of stock quotes (user interfaces on the traders’ desks)
subscribe and to which producers of stock quotes (so-called data
collectors) transmit.

The satellite dishes denote the equipment necessary to receive

financial events from a data provider, such as Reuters. The bar charts
depict graphical user interfaces residing on traders’ desks.
✓ In this example there are two channels, one for carrying Apple Computer stock quotes and
one for Sun Microsystems quotes.
✓ An assumption is that events are transmitted with a real multicast protocol such as IP
multicast. The network load is then constant, no matter how many listeners are active.

6-4 Distributed Programming With Java Technology

Object Bus Compared to CORBA and RMI

There are a few fundamental differences between Object Bus, and

CORBA and RMI. First, Object Bus systems are based on a group
communication model where talkers transmit events to a dynamically
growing and shrinking group of listeners. The logic necessary to
support registration and unregistration of listeners is transparent to
the developer of Object Bus applications.

In CORBA and RMI, a server object installs its object reference in a

name server. Clients retrieve that reference from the name server to
invoke the server. The Object Bus model does not make use of any
object references. Channels are typically named using character
strings, such as /dataproviders/financial/stockquotes/SUN.
Applications plug into a channel simply by specifying its name.

Object Bus Systems 6-5

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
6
Events are self-describing and self-contained. Typically, Object Bus
systems do not provide any interface definition language (IDL). A
talker can change the format of an event at runtime. However, the
listeners must be prepared to cope with format changes and reject any
event that does not provide the information needed by the listener.

The Object Bus model provides the abstraction of a ubiquitous

software medium extending through a whole company. Applications
“tap” into the medium at any location to send and receive events.
✓ A disadvantage is that type checking of event objects cannot be done at compile time any
more: in complex object bus applications the listeners typically need to cope with event
format changes due to software upgrades and the like.
✓ An advantage is that Object Bus systems can be extended at runtime since they do not
depend on static invocation stubs, as is the case in CORBA and RMI. Object Bus systems
are thus better suited to support a 7 days/24 hours per day of continuous operation since
talkers and listeners can be added (and even migrated) at runtime.

6-6 Distributed Programming With Java Technology

Object Bus Architectures

Object Buses are implemented in two different ways: hub-and-spoke

and multicast bus.

Hub-and-Spoke Architecture
The straightforward approach provides a network-centric “hub”
daemon. Listener applications connect to the hub; talker applications
use the hub to deliver an event on their behalf. There might be a hub
per channel, or a hub serving multiple channels simultaneously.
✓ Hub-and-spoke corresponds to the RMI broker architecture outlined in the “Relevance”
section.

The advantage of the hub-and-spoke architecture is that it is

straightforward to implement with RMI, CORBA, or plain TCP/IP
sockets. Another advantage is that you can use the hub for accounting
and access control purposes, because all events and subscriptions pass
through the hub.

Object Bus Systems 6-7

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
6
The disadvantage of hub-and-spoke is that each quote results in (at
least) one RMI request from the talker to the daemon, and in one RMI
per registered listener, to deliver the quote. RMIs are performed
whenever a listener registers or unregisters. This solution is unsuitable
in real-world scenarios where hundreds of different channels of
information are to be supported, each carrying real-time events to
many listeners. Furthermore, the hub embodies a single point of
failure.
✓ As was pointed out in the “Relevance” section, hub-and-spoke can also be implemented
by having the listeners poll the talker at regular time intervals, to check whether there are
any new events. This approach does not require any hub but consumes even more
network resources.

6-8 Distributed Programming With Java Technology

Multicast Bus Architecture

A more scalable solution consists of deploying a network multicast
protocol, such as IP multicast. This approach is more easily scalable,
because with IP multicast, the network traffic is independent of the
number of listeners subscribed to a channel.
✓ IP multicast deploys hardware-enabled multicast which is available in Ethernet and Token
Ring.

The disadvantage of the multicast bus architecture is that such an

Object Bus is more difficult to implement. Today’s operating systems
support only unreliable multicast, where messages can get lost on
overload or on network problems.

The advantage is that IP multicast, due to its fully distributed and

ubiquitous architecture, ideally supports the Object Bus model. In
addition, more efficient usage of network resources is ensured.
✓ IP multicast is well supported by Windows NT, Windows 95/98, all major UNIX environment
types, Linux, MacOS, and so on.

Object Bus Systems 6-9

✓ A special range of IP addresses (the class D) has been reserved for IP multicast. The
difference from conventional IP addresses is that applications running in an Intranet can
all bind to the same class D address and receive IP packets sent to the address. IP
multicast addresses thus are location independent.
✓ IP multicast is an unreliable protocol. Messages can get lost or duplicated. It is up to the
Object Bus middleware to provide the necessary reliability protocols. There is no TCP
over IP multicast. One typically creates UDP sockets to communicate through IP multicast.

6-10 Distributed Programming With Java Technology

Application Areas Suitable for Object Bus

The Object Bus model is well suited for applications that require one-
to-many or many-to-many transmission of real-time events to a
dynamically changing set of receivers. Examples are systems that push
financial events from a data collection unit to many trader
workstations. Teleconferencing applications and distributed
multimedia are other examples.

Another interesting application area is systems that need to be

extended with new functionality at runtime, without shutting the
system down. This is important for medical systems, flight reservation
systems, and other applications that require high availability.

Control systems for computer-integrated manufacturing (CIM) and

Workflow management are other examples.
✓ Workflow management systems are programmed to support the main work flows in a
company, such as a clerk taking a rental-car reservation by phone, checking a database
for a suitable car, assigning the car to the customer, and instructing the staff to prepare
the car.
✓ Numerous large banks have been deploying this kind of middleware for many years.

Object Bus Systems 6-11

Products and Standards

✓ Hints on the architecture of the product appear in parentheses. OrbixTalk is a pure C++
product. VisiBroker Event Service is available both for C++ and Java programming
language. Castanet is a pure Java technology middleware. TIBCO ObjectBus is
implemented in C and C++. iBus is a lightweight “pure Java” Object Bus.

The Object Management Group (OMG) Event Service standard defines

an Object Bus infrastructure on top of CORBA (www.omg.org). The
Event Service specification distinguishes event producers, event
consumers, and communication channels. A variety of
implementations of the standard are available today; for example,
OrbixTalk from Iona Ltd. (www.iona.com) and VisiBroker Event
Service from Inprise (www.inprise.com).

Marimba’s Castanet (www.marimba.com) is a Java technology Object

Bus targeted at distributing software upgrades and other data from
one source to many listeners.

TIBCO ObjectBus is a CORBA-based middleware in use at many

financial companies (www.tibco.com). SoftWired iBus is described on
the next overhead.

6-12 Distributed Programming With Java Technology

SoftWired iBus

The iBus Object Bus middleware by SoftWired Ltd.

(www.softwired.ch) is a pure Java technology implementation of the
multicast bus architecture. Its main characteristics are compact size,
easy installation, and an intuitive API.

iBus is designed so that it can run on top of various communication

protocols, notably reliable IP multicast and TCP. You can extend it
with new quality-of-service features, such as proprietary encryption
engines.

iBus provides an intuitive API allowing developers to subscribe to

communication channels and to transmit arbitrary Java objects using
those channels.
✓ iBus mainly consists of one JAR file. No naming services or broker daemons need to be
installed in order to run iBus applications. Since it is pure Java technology it can run on
any platform providing a Java runtime.

Object Bus Systems 6-13

The Talker Program

The following example provides a simple but complete iBus
application that transmits a stock quote on a channel carrying financial
information. Before any data can be transmitted, you must declare a
communication stack. In the constructor of class Stack, a quality-of-
service string is specified. Reliable multicast is used because the
example application requires one-to-many communication, and you
want to ensure that all quotes are received by all listeners, in spite of
packet loss, which might happen at the IP multicast level.
✓ iBus employs a negative-acknowledgments reliable multicast protocol but can be
configured for other protocols as well.

You must declare a URL for the channel to which you want to
transmit. registerTalker informs iBus that transmission will take
place using the stack and to the given channel.

Now it is time to create a posting object to fill with a quote for Sun
Microsystems shares. Transmit the posting by invoking the push
operation on the stack object. push is an asynchronous operation
meaning that it returns immediately without waiting until the listeners
have received the posting.
✓ iBus also provides a synchronous (that is, blocking) push operation as well as a
request/reply style operation, called pull.

The iBus protocol stack takes care of fragmenting large postings into
chunks fitting an IP datagram size, recovering from packet loss,
bringing packets in a first-in-first-out order, and so on. More
sophisticated stacks might perform failure detection, posting
encryption, and so on.

6-14 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
6
1 import iBus.iBusURL;
2 import iBus.Posting;
3 import iBus.Stack;
4
5 public class Talker {
6 public static void main(String [] argv)
7 throws Exception {
8 String quote = new String("SUN: 42.7");
9
10 // handle any command line argument:
11 if(argv.length > 0) quote = argv[0];
12
13 // create an iBus protocol stack for
14 // reliable multicast:
15 Stack stack = new Stack("Reliable");
16
17 //create an iBus URL for the destination
//channel:
18 iBusURL url = new iBusURL(
19 "ibus://226.1.2.3/financial/Text");
20
21 //open the channel:
22 stack.registerTalker(url);
23
24 //create a Posting to hold a quote string:
25 Posting posting = new Posting();
26 posting.setLength(1);
27 posting.setObject(0, quote);
28
29 //push the quote through the iBus channel
//"url":
30 for (;;) {
31 stack.push(url, posting);
32 Thread.currentThread().sleep(2000);
33 }
34 }
35 }

Object Bus Systems 6-15

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
6
The Listener Program
This application provides a listener that subscribes to the channel and
displays stock quotes. iBus implements an upcall event-handling
model in which iBus.Receiver objects are subscribed to with a
channel to receive postings. iBus invokes the
Receiver.dispatchPull method when a posting arrives on a
channel to which the receiver object is subscribed.

A class QuoteReceiver that implements interface iBus.Receiver is

defined. The dispatchPush operation extracts the quote string from
the posting and writes it to the output stream.

Now the main body of the listener creates an instance of

QuoteReceiver and subscribes it with the stack object and with the
channel. Finally, waitTillExit is called to tell iBus to suspend the
main thread and to wait for postings to arrive.
✓ Without calling waitTillExit the application would terminate immediately without
receiving any postings.

6-16 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
6
1 import iBus.iBusURL;
2 import iBus.Posting;
3 import iBus.Receiver;
4 import iBus.Stack;
5
6 public class Listener {
7 public static void main (String [] argv)
8 throws Exception {
9 Stack stack = new Stack("Reliable");
10 QuoteReceiver receiverObject =
11 new QuoteReceiver();
12 iBusURL url = new iBusURL(
13 "ibus://226.1.2.3/financial/Text");
14
15 stack.subscribe (url, receiverObject);
16 stack.waitTillExit();
17 }
18 }
19
20
21 // A listener class to receive quotes:
22 class QuoteReceiver implements iBus.Receiver {
23 public void dispatchPush(iBusURL source, Posting
p) {
24 // to display the quote string:
25 System.out.print("QuoteReceiver: got a quote:
");
26 System.out.println(p.getObject(0));
27 }
28
29 // not used in exercise
30 public Posting dispatchPull(iBusURL c, Posting p)
{
31 return null;
32 }
33 public void error(iBusURL channel, String details)
{}
34 };

Object Bus Systems 6-17

Exercise objective – Experience using an Object Bus.

Tasks

Compile and Run the Sample Application

Complete the following steps:

1. Change the directory to labfiles/mod6-objectbus/

lab/simple/.

2. Compile Talker.java and Listener.java.

3. Start the Listener program in a console window.

4. Start the Talker program in a console window.

5. Watch the output. How many Talker applications do you see?

✓ Each Talker in the network should show up immediately on every Listener. This may
surprise students and is a good demonstration for object bus technology.

Add and/or Remove Talkers and Listeners (Optional)

You can add more Listeners and Talkers now. You can shut a Talker
down, relocate it to another console window, for example, and restart
it without disturbing the Listeners.

6-18 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
6
Exercise Summary
Discussion – Take a few minutes to discuss what experiences, issues,
or discoveries you had during the lab exercises.

● Experiences
✓ Ask students what their overall experiences with this exercise have been. You may want to
go over any trouble spots or especially confusing areas at this time.

● Interpretations
✓ Ask students to interpret what they observed during any aspects of this exercise.

● Conclusions
✓ Have students articulate any conclusions they reached as a result of this exercise
experience.

● Applications
✓ Explore with students how they might apply what they learned in this exercise to
situations at their workplace.

Object Bus Systems 6-19

The iBus API

The Stack
iBus applications communicate through so-called communication
stacks. A stack defines a quality of service, such as reliable multicast,
real-time multicast, reliable point-to-point streaming, data encryption,
and so on. iBus is extensible in respect to incorporating yet
unsupported qualities of services.

6-20 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
6
The following is an example of the iBus API for sending and receiving
events.
1 package iBus;
2
3 public class Stack {
4 /** Create a protocol stack the application can use to send
5 * and receive Postings. Specify a quality of service such as
6 * reliable multicast or reliable streaming.
7 */
8 public Stack(String qos) { ... }
9
10 /** Register as a talker of a channel. This is to be done
11 * before any data can be pushed to the channel.
12 */
13 public void registerTalker(iBusURL channel) { ... }
14
15 /** Unregister as a talker.
16 */
17 public void unregisterTalker(iBusURL channel) { ... }
18
19 /** Subscribe a listener object to a channel.
20 */
21 public void subscribe(iBusURL channel, Receiver rcv) { ... }
22
23 /** Unsubscribe a listener object.
24 */
25 public void unsubscribe(iBusURL channel, Receiver rcv) { ... }
26
27 /** Push a posting through a channel. (one-way and asynchronous)
28 */
29 public void push(iBusURL channel, Posting p) { ... }
30
31 /** Pull a posting through a channel. Much like RMI.
32 */
33 public Posting[] pull(iBusURL channel, Posting request) { ... }
34 }

Object Bus Systems 6-21

Channels and URLs

When a Java talker application wants to communicate using iBus, it
must specify a transmission channel using a URL. iBus URLs obey the
following format:
ibus://address[:port]/subject

The address part contains a host name or an IP address. IP addresses

can be of the class A, B, C, or D.
✓ A, B, and C are “conventional” IP addresses used for point-to-point datagrams. Class D
addresses are multicast, these range from 224.0.0.0 to 239.255.255.255. This means there
is over a quarter of a billion possible IP multicast channel addresses!

A URL denotes a multicast or a point-to-point channel, depending on

its address part. The overhead on this page provides an example of a
multicast and a point-to-point URL.

6-22 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
6
Multicast channels offer more flexibility because one-to-many and
many-to-many communication is supported. Another advantage is
that such channels are not tied to particular computers but are truly
ubiquitous. This means that you can relocate iBus applications from
one machine to another in a “plug-and-play” fashion, without needing
to contact any central registry or naming service.

iBus multicast channels are used primarily within intranets. Secure

TCP bridges can be established to route iBus events from one intranet
to another.

Object Bus Systems 6-23

Posting Objects
Any Java object that is serializable can be transmitted using iBus. This
includes user-defined objects, JFC components, JavaBeans, and so on.
iBus provides the class Posting to help in packing several Java
objects and transmitting them at once.

6-24 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
6
1 T package iBus;
2
3 public class Posting {
4 /**
5 * Create a posting object to be transmitted through iBus
6 */
7 public Posting() { ... }
8
9 /**
10 * Pack an object into the posting.
11 */
12 public void setObject(int index, Serializable object) { ... }
13 /**
14 * Extract an object from the posting.
15 */
16 public Serializable getObject(int index) { ... }
17
18 /**
19 * Set the length property of the posting.
20 */
21 public void setLength(int length) { ... }
22
23 /**
24 * Get the length property of the posting.
25 */
26 public int getLength() { ... }
27
28 /**
29 * Get the iBus URL of the sender of the posting.
30 */
31 public iBusURL getSender() { ... }
32 }

Object Bus Systems 6-25

Exercise objective – Show how you can use objects more complex
than simple Strings in object bus applications.

Tasks

Build a Small Stock Quote Application

Use the Quote class to send and receive stock quotes in a extremely
simplified stock quote example. Complete the Talker and the Listener
and run the application.

1. Change the directory to labfiles/mod6-objectbus/lab/

quote.

2. Compile Quote.java.

3. Write a Talker that generates and sends stock quotes described by

instances of the Quote class. Use the skeleton
TalkerQuote.java.

4. Write a Listener that receives stock quotes and processes them

(displaying on the console). Use the skeleton
ListenerQuote.java.

5. Compile the Talker and the Listener and run the application.

6. As more and more people in your class complete the exercise and
run their Talkers, your Listener will automatically get all stock
quotes sent by all Talkers.

6-26 Distributed Programming With Java Technology

● Experiences
✓ Ask students what their overall experiences with this exercise have been. You may want to
go over any trouble spots or especially confusing areas at this time.

● Interpretations
✓ Ask students to interpret what they observed during any aspects of this exercise.

● Conclusions
✓ Have students articulate any conclusions they reached as a result of this exercise
experience.

● Applications
✓ Explore with students how they might apply what they learned in this exercise to
situations at their workplace.

Object Bus Systems 6-27

Before continuing on to the next module, check that you were able to
accomplish the following:

❑ Understand the fundamentals of the Object Bus model

❑ Understand its main application areas

❑ Explain the differences between iBus, RMI, and CORBA

6-28 Distributed Programming With Java Technology

What benefits are there in using an Object Bus middleware for

building a mission-critical system, such as financial data feed?
✓ Extensibility. Add further listeners and talkers to a running system and add new channels
to a running system.
✓ Simple architecture. Components need to interface only to the bus. This makes for a “flat”
architecture since components do not reference each other directly.
✓ Instant delivery of events.

Object Bus Systems 6-29

Objectives
Upon completion of this module, you should be able to:

● Describe the Java Naming and Directory Interface

● Write a simple JNDI client that binds a CORBA object into the
service and later looks it up

● Describe the Java Transaction Service

● Describe the Java Message Service

No matter which technology you choose to distribute your

application, you will almost always need one or more supporting
technologies: a naming service to look up addresses or objects by
name; a transaction service to manage distributed consistency; or a
messaging service to let your components communicate
asynchronously.

Discussion – Consider the following questions:

● If all you know about a remote object is its name, how do you get
its reference?

✓ So far, a technology-specific naming service, rmiregistry with RMI and COS naming with
JavaIDL, has been used.

● Would it be preferable to have the same API for all available

naming services?
✓ There is no standard answer to this question. If you already know the APIs to several
naming services, why learn yet another one? On the other hand, if there is just one API,
there is less to learn if you don’t know any naming service APIs yet.

● How would you design a vendor-independent transaction API?

✓ Two possibilities come to mind: either use an already common and widely used standard
API, or design a new one. The Java Transaction Service does both: It supports the CORBA
COS Object Transaction Service, and it supports a new, Java technology-specific
transaction API.

● What do you think about when you hear the term messaging?
✓ Often, email comes to mind. However, messaging is not email. While messaging shares
some characteristics with email (especially the asynchronous nature of the service),
messaging is about asynchronous requests, reports, or events that are consumed by
enterprise applications, not humans.

7-2 Distributed Programming With Java Technology

Overview

This module looks at three supporting technologies from JavaSoft:

Java Naming and Directory Interface (JNDI), Java Transaction Service
(JTS), and Java Message Service (JMS).

Supporting Technologies 7-3

Java Naming and Directory Interface

In today’s information technology (IT) world, there is a wide range of

naming and directory technologies and products available. The
following lists some of these products:

● Domain Name System (DNS)

● Network Information Services (NIS and NIS+)

● Lightweight Directory Access Protocol (LDAP)

● CORBA naming service (COS naming)

● RMI naming service (rmiregistry)

● The OSI protocol for managing online directories of users and

resources (X.500)

● NetWare Directory Service (NDS)

7-4 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
7
Each of these products has its API (mostly for C or C++). To access
these products from a Java application, a Java API is necessary. There
are a couple of Java technology APIs; Netscape developed a Java
technology API to LDAP, you can reach DNS using the java.net
package, COS naming is part of the Java 2 SDK, and RMI is a Java
technology. However, to reach all the products from Java applications,
more APIs are necessary.

JNDI takes another approach. It provides a unified Java API to any of

the naming and directory services available. The JNDI API can even be
used if another Java API already exists; you can reach COS naming,
LDAP, DNS or the rmiregistry using JNDI.

A developer now faces this choice: use JNDI, or use the “older” API?
How you decide depends on whether you know the “older” API. If
you know how to reach COS naming, there is no advantage in using
JNDI. On the other hand, once you know JNDI, you can use this
knowledge to reach other naming and directory services easily.

Supporting Technologies 7-5

Naming Services and Directory Services

Naming and directory services are often confused, but the following
sections describe the differences.

Naming Services
A naming service does only one task: it looks up an object, given the
object’s name. You have to know the name of the object you are
looking for, or you will not find anything. The types of objects that are
supported depends on the particular naming service. DNS, for
example, supports only IP addresses; COS naming service supports
CORBA objects.

7-6 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
7
Directory Services
A directory service adds the capability for searching. You no longer
have to know the name of the object for which you are looking.
Instead, you can provide a set of attributes that your desired object
should have set.

In a real-world example, you have to know the name to find a person’s

phone number and address. This is insufficient if you want the address
of all people who like skiing, for example. If you can find a company
who has a phone book on disk, and has an attribute “hobby” attached
to each person, you might get what you want. This company would
provide a directory service, extending the possibilities of a name
service alone.

For more information about the directory service part of JNDI, consult
the Java Naming and Directory Interface specification
(http://java.sun.com/products/jndi/ ).

Supporting Technologies 7-7

Composite Names

Usually, you do not have one large dictionary that contains all the
names. Instead, the dictionary is distributed in a hierarchical manner.
The domain name system, for example, is highly distributed; each
company manages the machines in its own domain. The domains, in
turn, are managed differently for example, by a public organization
and grouped to a country-wide context. To reach a domain in another
country, still higher hierarchical levels are needed. A single dictionary
would be a namespace.

Namespaces are usually linked; in the address java.sun.com there is

the com namespace, which points to the sun namespace. You can use
the sun namespace to look up the address of the java host.

Namespaces can even be a combination of several name services. The

address java.sun.com/products/jndi/docs.html is a combination
of DNS (java.sun.com) and a local file system
(products/jndi/docs.html). JNDI can manage namespaces with an
arbitrary number of different services involved.

7-8 Distributed Programming With Java Technology

Architecture Overview

The Java Naming and Directory Interface specification declares two

APIs: the JNDI API, and the JNDI Service Provider Interface.

The JNDI API is used by developers to get at the services of a naming

or directory service. A part of this API will be used later in the
exercises.

The JNDI Service Provider API is used to link a specific naming or

directory service to the JNDI system. A module that plugs into JNDI
using the Service Provider API is called a service provider.

Supporting Technologies 7-9

Service Providers

The goal of JNDI is to eventually reach all available naming and

directory services using JNDI. To reach this goal, service providers,
which plug using the Service Provider API to JNDI, must be
developed. The following service providers are available already:

● COS naming

● LDAP

● NIS, NIS+

● Novell NDS

● File system

● rmiregistry

7-10 Distributed Programming With Java Technology

Module Exercise Overview

Most of the services are not available or not configured correctly in a

classroom. However, there are still two to choose from: COS naming,
which is part of the Java 2 SDK, and the local file system. Of the two,
COS naming is more interesting, because there is already an
application that uses it: the IDL version of Echo.

In Module 4, the client and the server both used COS naming directly
to bind and look up the Echo implementation. If you convert to use
JNDI, the picture looks different; the client uses JNDI to look up the
object reference. This is translated using the COS naming service
provider to COS naming. Because it is translated on the client, the
server can continue to use COS naming directly.

According to the previous description, it should also be possible to

convert the server to JNDI, and use either the “old” client (which uses
COS naming directly) or the JNDI version of the client, to reach the
JNDI-based server.

Supporting Technologies 7-11

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
7
Using JNDI to Access COS Naming
Using the Java Naming and Directory Interface is straightforward. The
lookup of an object consists of three steps:

1. Set the necessary properties.

2. Get a first Context.

3. Use the Context to look up either of the following:

▼ Another Context further down the hierarchy

▼ The desired object.

Of course, step 3 might be repeated several times.

You use properties to configure the JNDI core and the individual
service providers. As usual, you can set a property in the form of a
system property (using the -D command-line option), or by passing a
Properties object on JNDI initialization. Because the class
java.util.Properties accepts only Strings, but some service
providers require arbitrary objects to configure themselves, you can
use a java.util.Hashtable instead of a Properties object. Setting
such a property with the -D command-line option does not work.

The following is the necessary code to look up a CORBA object:

Hashtable env = new Hashtable(5, 0.75f);
env.put("java.naming.factory.initial",
"com.sun.jndi.CosNaming.CNCtxFactory");
env.put("java.naming.corba.orb", orb);
Context ic = new InitialContext(env);
Echo echoRef = EchoHelper.narrow(
(org.omg.CORBA.Object)ic.lookup("Echo"));

The first of the two properties that are set,

java.naming.factory.initial, is used by the JNDI core to generate
the initial context object. The second property,
java.naming.corba.orb, is used by the COS naming service
provider to get a reference to the ORB.

7-12 Distributed Programming With Java Technology

Exercise objective – Learn about JNDI and the COS naming service
provider by coding an example.

Preparation
Install the JNDI JAR file, and two of the service providers: COS
naming and local file system.

Thanks to the new extensions framework of the Java 2 SDK, the

installation is easy. Copy the JAR files to the $JAVA_HOME/lib/ext
directory. There is no need to update the CLASSPATH.

Tasks

Modify the JavaIDL Echo Server to Use JNDI Instead of COS

Naming

So far, in all the JavaIDL examples, the CORBA native naming service,
COS naming, was used. Thanks to the COS naming service provider,
you can use the Java Naming and Directory Interface instead. A
modified client is already provided as part of this exercise. Your job is
to modify the server as well, and then to test whether “old” and “new”
clients and servers can interoperate.

1. Change the directory to labfiles/mod7-support/lab/jndi.

2. Analyze the code of the modified client, which uses JNDI instead
of COS naming to get the remote Echo object reference.

3. Compile and run the application. Can the modified client

interoperate with the current server?

Supporting Technologies 7-13

4. Modify the server to use JNDI as well.

5. Try the different combinations of COS naming or JNDI on the

server and on the client. Do COS naming and JNDI with COS
naming service provider really work together transparently?

Can You Use the File System Service Provider Instead of COS
Naming?

The Java Naming and Directory Interface unifies the API to different
products. Does it also mask the implementation differences? You can
find out by trying to use the file system service provider instead of the
COS naming service provider.

1. Start with the server code, and replace the COS naming service
provider with the file system service provider. The initial factory
is com.sun.jndi.fscontext.FSContextFactory. Instead of a
handle to the ORB, the file system service provider needs a URL-
like string that indicates the root of the file system. The property
called java.naming.provider.url can be set, for example, to
file:///tmp, or file:///D:/temp

If it does not work, seek an explanation and discuss it with the

other students.
✓ The file system service provider does not accept CORBA objects in a bind or rebind call.
If you think about this for a second, it becomes clear why. Even with “pure” COS naming,
the object to bind on the server and the object to look up on the client is not the same.
You bind an “EchoImpl” to the COS naming service, but get a stub out of COS naming on
the client. This functionality is specific to the COS naming service and is, of course, not
implemented in the file system service provider. Even if you could bind arbitrary objects
to the file system, you would get the wrong object out of it on the client. (You would get
an EchoImpl instead of a stub.)

7-14 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
7
Exercise Summary
Discussion – Take a few minutes to discuss what experiences, issues,
or discoveries you had during the lab exercises.

● Experiences
✓ Ask students what their overall experiences with this exercise have been. You may want to
go over any trouble spots or especially confusing areas at this time.

● Interpretations
✓ Ask students to interpret what they observed during any aspects of this exercise.

● Conclusions
✓ Have students articulate any conclusions they reached as a result of this exercise
experience.

● Applications
✓ Explore with students how they might apply what they learned in this exercise to
situations at their workplace.

Supporting Technologies 7-15

Java Transaction Service

The Java Transaction Service (JTS) defines a standard transaction

management API for the Java platform. JTS consists of the following
elements:

● A standard Java technology mapping (“Java mapping”) of the

OMG Object Transaction Service (OTS). The Java mapping of the
OTS interfaces is defined in the packages
org.omg.CosTransactions and org.omg.CosTSPortability.

● An application-level transaction demarcation API. This API is

defined in the package javax.jts.

● A Java mapping of the industry-standard XA interface. This API is

defined in the javax.jts.xa package. This package is a Java
binding of the X/Open XA interface. The XA interface defines the
API through which a transactional resource manager attaches to
an external transaction manager.

7-16 Distributed Programming With Java Technology

Java Message Service

Just as Java Naming and Directory Service provides a unified interface

to existing naming and directory products, the Java Messaging Service
(JMS) provides a generalized API to various enterprise messaging
service products. JMS provides a common way for Java programs to
create, send, receive, and read an enterprise messaging system’s
messages.

JMS is a set of interfaces and associated semantics that defines how a

JMS client accesses the facilities of an enterprise messaging product.
While you could write and compile a client application, it would not
run with JMS alone, you need at least one service provider. Whether
the final version of JMS will include a simple service provider for
testing purposes is not known at this time.

JMS Service Provider

As with JNDI, a JMS provider maps the JMS API to a specific
messaging product. Ideally, JMS providers will be written in 100%
Pure Java™ so they can run in applets, simplify installation, and work

Supporting Technologies 7-17

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
7
across architectures and operating systems. However, early
implementations will probably provide only a JMS API on top of a
legacy product.

JMS Domains
There are two different types of messaging products on the market
today. One group of products uses point-to-point semantics; that is, a
message is sent from one sender to one receiver. The other group uses
publish and subscribe semantics, which is typically a many-to-many
approach. The JMS API provides support for both types of messaging
products.

7-18 Distributed Programming With Java Technology

JMS Domains – Point-to-Point

Point-to-point (PTP) products are built around the concept of message

queues. Each message is addressed to a specific queue and clients
extract messages from the queue(s) established to hold their messages.

Supporting Technologies 7-19

JMS Domains – Publish and Subscribe

Publish and subscribe (or “publish-subscribe”) clients address

messages to some node in a content hierarchy. Publishers and
subscribers are generally anonymous and can dynamically publish or
subscribe to the content hierarchy. The system takes care of
distributing the messages arriving from a node’s multiple publishers
to its multiple subscribers.

7-20 Distributed Programming With Java Technology

Before continuing on to the next module, check that you were able to
accomplish the following:

❑ Describe the Java Naming and Directory Interface

❑ Write a simple JNDI client that binds a CORBA object into the
service and later looks it up

❑ Describe the Java Transaction Service

❑ Describe the Java Message Service

Supporting Technologies 7-21

How can the supporting technologies help in your development of a

distributed system?

7-22 Distributed Programming With Java Technology

Objectives
Upon completion of this module, you should be able to:

● Explain the role of JDBC and servlets in a distributed application

● Compare and contrast RMI and JavaIDL

● Compare and contrast request-reply and publish-subscribe

architectures

This module summarizes and compares the technologies presented in

this course. It will help you decide which technology is appropriate to
use in different circumstances.

Discussion – Consider the following questions:

● Will you be using all the technologies you learned about in one
and the same application?

✓ Usually not. RMI and JavaIDL tend to exclude each other. This is not a technical issue, but
one of personal preference, company regulations, or existing legacy code.

● Is JDBC always a part of a distributed application?

✓ Often it is, because in real-world applications there is usually a persistent data store
around; that is, a database. However, if there is no database, there will be no JDBC.

8-2 Distributed Programming With Java Technology

Complementary or Overlapping Technology?

This course describes five distributed computing technologies:

● JDBC for database connectivity

● Servlets for extending a Java technology-based server

● RMI to create and use remote objects

● JavaIDL to create and use remote objects

● Object Bus to support mobile objects and near realtime object

delivery

From these five technologies, only two overlap to a large degree: RMI
and JavaIDL. Both technologies serve basically the same purpose: they
allow for the development of remote objects. To a lesser degree
RMI/JavaIDL and object buses overlap, because you can solve several
distributed computing problems using either technology.

Technology Summary and Comparison 8-3

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
8
In this module, the comparison is therefore between RMI and JavaIDL
on one hand, and between request/reply style communication (for
example, RMI or JavaIDL) and publish/subscribe style communication
(for example, object buses) on the other hand.

Additionally, this module describes the roles of JDBC and servlets in a

distributed computing environment.

8-4 Distributed Programming With Java Technology

JDBC Usage

JDBC is used for one simple purpose: to work with databases. In the
Java technology environment, there is no alternative to JDBC. There
are commercial tools to assist the developer, such as Java Blend™ from
Sun Microsystems. But even this tool produces JDBC code as a result.

You can use JDBC directly on a client system, although it is not the
preferred usage. Mostly JDBC is used on a business logic server, which
translates the raw JDBC calls into a higher-level business object. This
object is then called remotely, using RMI or JavaIDL.

Technology Summary and Comparison 8-5

Using Servlets

The application area of servlets is broader than that of JDBC. In a

generic sense, servlets are used to extend the functionality of any
server based on the Java programming language. However, the
primary usage of servlets today is extending a Web server.

Because servlets are written in the Java programming language, they

can use JDBC, RMI, or JavaIDL.

8-6 Distributed Programming With Java Technology

Example of an n-Tier Architecture

The above overhead shows the picture of a simple n-tier architecture.

The core is a business logic server, running the UNIX operating
system. It uses JDBC to access data residing on two different legacy
hosts.

A client capable of running Java programs can access the business

logic server directly, using RMI or JavaIDL. A pure HTML client needs
a helper to present the data, for example, in an HTML form. This
helper is written as a servlet. This servlet communicates with the
business logic server using the same remote objects as the client.

Technology Summary and Comparison 8-7

RMI Compared to JavaIDL

The Java 2 platform allows for the building of distributed object

applications using two different methods: RMI and JavaIDL. Both
technologies have their advantages and disadvantages, which should
be compared and contrasted to solve the confusion of having two
apparently overlapping solutions.

There are three types of Java technology developers:

● CORBA veterans have developed CORBA applications using C++

or other languages. The Java technology environment, for them,
promises to take CORBA to the place it belongs. RMI, in contrast,
is completely unnecessary for them and should be ignored.

● For Java technology purists, all applications must be written or

rewritten in the Java programming language. If all components are
developed with the Java programming language, why use a
technique that is made for the support of different languages?

● The third group consists of all the developers that are new to
distributed objects in the Java technology environment, and are
not prejudiced towards a certain solution.

8-8 Distributed Programming With Java Technology

JavaIDL Advantages

There are two kinds of advantages of the CORBA architecture:

● Technical advantages

● Market advantages

Technical advantages include the fact that CORBA was designed

without a special language in mind. This gives advantages as well as
disadvantages; however, if there is legacy code or code written in a
language other than the Java programming language, CORBA is
probably the superior solution for a given project. Interfaces between
clients and servers are defined in IDL, providing the advantages of
multi-language and multi-platform environments and enforcing a
clear separation between interfaces and implementations. Another fact
is the rich set of standardized services that are useful for distributed
object applications, such as the Lifecycle, Naming, or Event Service.

Another technical advantage is the support for dynamic method

discovery and invocation.

Technology Summary and Comparison 8-9

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
8
It is significant that the CORBA architecture is supported by a
consortium of over 800 companies, representing the whole variety of
the computer industry. They are willing to make this architecture the
next generation of middleware. The results of this support are the
many product offerings and various commercial alternatives from
which customers can choose.

8-10 Distributed Programming With Java Technology

JavaIDL Disadvantages

Current CORBA implementations do not allow the passing over an

object and its current state to the remote system (objects by value).
However, there is already a draft standard to support this, which soon
will be followed by implementations.

The CORBA architecture is made for heterogeneous systems. Its goal is

to make all languages cooperate, so that arguments can easily be
passed over. To achieve this goal, a lowest common denominator
approach is required. You cannot pass a Java object to a COBOL
program, because the COBOL program cannot “know” what to do
with it. Instead, objects are mapped to IDL, and implemented for
every target language and platform. This approach provides a certain
level of interoperability, while losing the power of a specific language
environment.

Technology Summary and Comparison 8-11

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
8
RMI, on the other hand, can take a different stance; it can make the
assumption that it deals with only one language environment on both
sides of the wire, leveraging all the inherent advantages of the Java
object environment. CORBA cannot allow a type-safe object passing,
because it cannot make any assumptions about which language is
used on the other side. With RMI, it is even possible to pass a subclass
of some object to a remote VM, and the remote VM will then
download the code, and invoke methods of the subclass
polymorphically from the base classes interface. With CORBA it is
mandatory that both sides know in advance exactly what object will
be passed.

A Java applet that wants to talk to a remote CORBA object needs an

ORB to communicate with. Therefore, a so-called orblet must be
downloaded to the client together with the applet code. This happens
each time the client tries to start the applet. While the common
bootstrapping protocol might help on this issue, not every ORB has
implemented locally all of the potentially required CORBAservices,
like the full Security or Trading Service.

8-12 Distributed Programming With Java Technology

Java RMI Advantages

Java RMI is tightly integrated into the Java technology environment

and once the remote object is located, you can invoke methods as if it
were local.

RMI also uses an optimized protocol for the communication between

remote Java objects. Data marshalling and object serialization are
handled transparently for the developer.

With RMI, Java objects in different VMs can transparently invoke each
other’s methods. Since VMs can exist on different systems, RMI can
provide full polymorphism.

The RMI system enables the garbage collection of active objects. The
distributed garbage collection mechanism frees the developer of all the
tedious memory management.

RMI allows behavior to be moved between VMs. Java classes can be

built into clients and then sent to the server or vice versa. Passing Java
objects in this way preserves the encapsulation. Unlike CORBA, RMI
allows you to pass objects by reference as well as by value.

Technology Summary and Comparison 8-13

Java RMI Disadvantages

The 100% Pure Java RMI solution has an obvious drawback: the
interaction with objects written in different languages is not possible.
RMI does not provide methods to use services offered by C++ or
Smalltalk objects. The integration of legacy code is a major problem for
pure RMI applications.

Today CORBA offers services that are not available in RMI. There is,
for example, no event support in RMI; a Java applet waiting for a
specific event notification has to poll its server frequently to get the
updates. The RMI naming service today is non-persistent and quite
primitive. If an object is changed, the rmiregistry daemon has to be
stopped, and the whole RMI subsystem must be restarted.

RMI is implemented on TCP/IP instead of IIOP. Most firewalls refuse

TCP/IP communication, while others like Iona’s wonderwall at least
allow IIOP through firewalls. RMI tries to achieve the same effect by
tunnelling its communication in HTTP, in this way disguising
communication as HTTP POSTs and GETs.

8-14 Distributed Programming With Java Technology

RMI Over IIOP

JavaSoft and OMG are currently working on a convergence of the

CORBA and RMI object models. This concerns two areas:

● RMI over IIOP. With the release of the Java 2 SDK comes the
support for RMI that to work on top of IIOP, thus providing the
following benefits to RMI:

▼ Built-in transaction support, using OMG OTS

▼ Interoperability with objects written in different languages via

the RMI/IDL subset

● RMI/IDL. With the CORBA Java-to-IDL standard, it is possible to

define CORBA interfaces with Java RMI semantics. The RMI/IDL
allows CORBA clients to call RMI objects and vice versa. RMI
clients can therefore invoke methods on CORBA objects written in
different languages.

Technology Summary and Comparison 8-15

RMI and JavaIDL

The table in the above overhead summarizes the features of RMI and
JavaIDL. This course shows that the processes of building applications
using these technologies are fairly similar. Each has inherent
advantages the other does not match. Java RMI’s advantage of being
built right into the Java programming language, allowing it to exploit
the full capabilities of the Java programming language will never exist
in JavaIDL. The JavaIDL, though, has the advantage of working with
existing legacy code.

8-16 Distributed Programming With Java Technology

Request-Reply Compared to Publish-Subscribe

Request-reply is what most developers grew up with. A request (that is,

a function call with arguments) is sent to a well-known service
provider (that is, a remote object), and a reply (a return value) is sent
back. Such a conversation is inherently point to point. An example of
request-reply is shown on the above overhead. A person requests a
stock quote, and receives the answer later. If several people want the
stock quote, they all have to ask individually.

An example of publish-subscribe is also shown on the above overhead.

Several people gather around a specific loudspeaker (they “subscribe”
to whatever the speaker has to say). Somebody then talks into the
microphone (something is “published”) and announces a stock quote.
Everybody around the speaker can hear the quote at the same time: at
the precise moment the quote is announced. Those not near the
speaker will not hear the announcement.

Technology Summary and Comparison 8-17

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
8
Another difference from request-reply is the anonymity. The
subscribers need to know only the position of the loudspeaker (the
“channel name” in software bus terminology), not the person talking
into the microphone (the remote reference of the object that is
publishing).

Since most people are familiar with request-reply, but not with
publish-subscribe, two publish-subscribe scenarios are shown on the
next two overheads. The first scenario is an application that is clearly
intended for a bus. The second scenario is an application that could be
solved (and typically is) with request-reply. This would be a case
where request-reply and publish-subscribe overlap.

8-18 Distributed Programming With Java Technology

Bus Example 1 – Real-Time Enterprise

Today’s enterprises need specific information at near-real time. Not

everybody, however, needs the same information. Sending mass email
is inefficient and not easy to administer. A publish-subscribe medium
can help in this scenario.

On the above overhead, four different channels are shown: one

carrying stock quotes, one carrying selected usenet-news, one carrying
alerts generated by intranet hardware, and one carrying inventory
changes.

The client application can be seen as a versatile tuner, which allows the
user to select relevant channels for display. More function-rich clients
are possible as well; for example, the hardware alerts could be
processed in a network management application.

Technology Summary and Comparison 8-19

Bus Example 2 – Flat Architecture

The above overhead shows the components of a simplified but typical

enterprise environment. Usually, the components are linked together
individually: order entry is linked with the customer database, with
order scheduling, with accounts receivable, and so on. These
connections are point to point (via CORBA or RMI) and hardcoded. If
the organization is changed, the system must be rebuilt.

On the overhead, the components are linked only to the bus. The order
entry components publish all the information about a newly entered
order onto a specific channel on the bus. Every component that needs
to know about orders is registered to this channel, and acts
accordingly. New functionality can be added by creating new
components and subscribing them to relevant channels.
Organizational changes can usually be dealt with by updating a few
components only. Other non-affected components never know about
the change, since the channel continues to deliver the expected
information.

8-20 Distributed Programming With Java Technology

Before continuing, check that you are able to accomplish the following:

● Explain the role of JDBC and servlets in a distributed application

● Compare and contrast RMI and JavaIDL

● Compare and contrast request-reply and publish-subscribe

architectures

Technology Summary and Comparison 8-21

How do all the distributed technologies work together?

8-22 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
Flags for the idltojavaUtility A
This appendix describes the options and environmental files
for the idltojava utility.

The idltojava utility compiles IDL files to Java source code.

Syntax
idltojava [ options | flags ] filename.idl ...

Description
The idltojava utility compiles IDL source code into Java source code.
You then use the javac compiler to compile that source to Java
bytecodes. The IDL declarations from the named IDL files are
translated to Java programming language declarations according to
the mapping from IDL to the Java programming language.

Options
Options are used to pass some environment-specific information to the
idltojava utility. The following three options deal with preprocessor
directives:

● -I directoryName – Deals with #include

● -D symbol – Deals with #define

● -U symbol – Deals with #undefine

All three tell the idltojava utility to pass on information to the

preprocessor.

A-2 Distributed Programming With Java Technology

● -j javaDirectory – Specifies that generated Java files should be

written to the given javaDirectory. There must be a space
between -j and the directory name. Also, you must create the
directory before specifying it in this option.

● -I directoryName – Tells idltojava to pass on to the

preprocessor the information that the directory directoryName
is the place to search for the files in #include directives contained
in the IDL file.

● -D symbol – Specifies that symbol must be defined during

preprocessing of the IDL file(s).

● -U symbol – Specifies that symbol must be undefined during

preprocessing of the IDL files.

Flags
Flags give instructions to the idltojava utility and can be turned on
or off. To turn on a flag, type the following:
idltojava -fflag-name.

For example, the following directs idltojava to print a list giving the
current state of each flag:
idltojava -flist-flags hello.idl

To turn off a flag, type the following:

idltojava -fno-flag-name

The following line directs idltojava not to use the C/C++

preprocessor before compiling the IDL file hello.idl:
idltojava -fno-cpp hello.idl

Each flag is set to a default value. Turning a flag off when it is already
off does nothing; similarly, turning a flag on when it is already turned
on does nothing.

Flags for the idltojava Utility A-3

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
A
If you run idltojava on an IDL file and supply no flags, idltojava
runs the C/C++ preprocessor on the IDL file, produces a portable
client stub file, produces a portable server skeleton file, considers case
when comparing identifiers, and writes out the .java files it produces.

The following are the flags you can use with idltojava:

● -fcaseles – When ON, requests that capitalization not be

considered in the comparison of identifiers. The default is OFF. In
other words, the default behavior of the idltojava compiler is to
consider case when comparing identifiers. Use -fno-
caseless.ogm to tell idltojava not to consider case.

● -fclien – When ON, requests the generation of the client side of

the IDL files supplied. Default is ON.

● -fcp – When ON, runs the C/C++ preprocessor on the IDL file
supplied. Default is ON.

Note – idltojava is hard-coded to use a default preprocessor. It uses

the path /usr/ccs/lib/cpp when looking for the preprocessor. You
can change the preprocessor that idltojava uses by setting two
environment variables:

CPP – Set this environment variable to the full path name of the
preprocessor executable you want to use.

CPPARGS – Set this environment variable to the complete list of

arguments to be passed to the preprocessor. The preprocessor needs to
write to standard output, so if it does not do so by default, include the
argument appropriate to your preprocessor to accomplish that.

-f list-flag – When ON, requests that the current state of all the
-f flags be printed. Default is OFF.

● -f list-option – When ON, requests a list of command-line

options. Default is OFF.

● -f map-included-file – When ON, requests that Java code be

generated for the IDL file and all the files listed in #include
directives in the IDL file. Default is OFF.

● -f portabl – When ON, requests the generation of portable stubs

and skeletons. Default is ON.

A-4 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
A
● -f serve – When ON, requests the generation of the server side of
the IDL files supplied. Default is ON.

● -f t – When ON, requests the generation of the Operations

interface and Tie class, which are required for the implementation
of a delegation-based skeleton. For the file hello.idl, idltojava
creates the files _helloTie.java and _helloOperations.java.
This flag does nothing if the -fserver flag is turned off. Default is
OFF.

● -f verbos – When ON, requests that idltojava print comments

on the progress of the compilation. Default is OFF.

● -f versio – When ON, requests that idltojava print its version

and timestamp. Default is OFF.

● -f write-file – When ON, requests that the Java files that are
generated be written out. idltojava creates a directory named
after the module for the .idl file and puts the .java files in it.
This new directory is in the same directory as the .idl file.
Turning the flag off allows the programmer to validate the IDL
code before writing out the generated Java files. Default is ON.

Using the #pragma Compiler Directive

Java IDL supports the #pragma compiler directive. A #pragma
directive must appear at the beginning of an IDL file so that it has
global scope.

To request a repository prefix, type the following:

#pragma prefix "requested_prefix_name"

To wrap the default package in one called package, type the

following:
#pragma javaPackage "package"

For example, compiling an IDL module M normally creates a Java

package M. If the module declaration is preceded by the directive
#pragma javaPackage "browser"

the compiler creates the package M inside package browser. This

#pragma is useful when the definitions in one IDL module are used in
multiple products.

Flags for the idltojava Utility A-5

The SQL standard specifies a set of commands and a specific syntax

for the retrieval and modification of data, as well as commands for the
administration of tables. Each SQL statement is issued to the database
system and parsed.

SQL statements begin with a command keyword. mSQL supports the

following set:

● SELECT – Retrieves zero or more records from a named table

● INSERT – Adds a new record to a named table

● DELETE – Removes one or more records from a table

● UPDATE – Modifies one or more fields of particular records

● CREATE – Builds a new table with the specified field names and
types

● DROP – Completely removes a table from the database

The syntax of these clauses is meant to be read like English commands

and can be spoken aloud. For example, an SQL command could be
“get me all of the fields in the table named Employee Data, where the
employee ID is 10223.” If the Employee Data table contained fields
that held a name, employee ID, date of hire, social security number,
and current salary, you would expect to receive a single employee
record with these values.

In mSQL, this command would be written using the following

statement syntax:
SELECT * FROM employee_data WHERE employee_id =
'10223'

Note – The employee_id field is presented here as a string, but if it

were an integer or a real number, you would not require the single
quotes.

mSQL statements are not case sensitive, but the examples shown here
highlight the keywords with capital letters.

B-2 Distributed Programming With Java Technology

Syntax
The SELECT statement is the primary command used for data retrieval
from a SQL database. It supports the following:

● Joins

● DISTINCT row selection

● ORDER BY clauses

● Regular expression matching

● Column-to-column comparisons in WHERE clauses

The formal syntax for mSQL’s SELECT is:

SELECT [table.]column [ , [table.]column ]...
FROM table [ , table]...
[ WHERE [table.]column OPERATOR VALUE
[ AND | OR [table.]column OPERATOR VALUE]... ]
[ ORDER BY [table.]column [DESC] [, [table.]column [DESC]
]

Where

● OPERATOR can be <, >, =, <=, >=, <>, or LIKE.

● VALUE can be a literal value or a column name.

The regular expression syntax supported by LIKE clauses is that of

standard SQL:

● An underscore (_) matches any single character

● A percent sign (%) matches zero or more characters of any value

● A back slash (\) escapes special characters (for example, \%

matches % and \\ matches \)

● All other characters match themselves

SQL Syntax B-3

To sort the returned data in ascending order by last_name and

descending order by first_name, use the following query:
SELECT first_name, last_name FROM emp_details
WHERE dept = 'finance'
ORDER BY last_name, first_name DESC

Note – Here DESC applies to both last_name and first_name.

To remove any duplicate rows, use the DISTINCT operator:

SELECT DISTINCT first_name, last_name FROM emp_details
WHERE dept = 'finance'
ORDER BY last_name, first_name DESC

Note – DISTINCT is useful when you do not have a primary key.

To search for anyone in the finance department whose last name

consists of a letter followed by “ughes,” such as “Hughes,” use the
following query:
SELECT first_name, last_name FROM emp_details
WHERE dept = 'finance' AND last_name LIKE '_ughes'

B-4 Distributed Programming With Java Technology

Note – The SQL JOIN keyword is not supported.

The power of a relational query language is apparent when you start

joining tables during a SELECT operation. For example, if you have
two tables defined—one containing staff details and another listing the
projects being worked on by each staff member—and each staff
member has been assigned an employee number that is unique to that
person, you can generate a sorted list of who was working on what
project with the following query:
SELECT emp_details.first_name, emp_details.last_name,
project_details.project
FROM emp_details, project_details
WHERE emp_details.emp_id = project_details.emp_id
ORDER BY emp_details.last_name, emp_details.first_name

mSQL places no restriction on the number of tables joined during a

query; if there are 15 tables, all containing information related to an
employee ID in some manner, data from each of those tables can be
extracted (albeit slowly) by a single query.

Note – You must qualify all column names with a table name. mSQL
does not support the concept of uniquely named columns spanning
multiple tables. You are forced to qualify every column name as soon
as you access more than one table in a single SELECT.

SQL Syntax B-5

Note – Unlike ANSI SQL, you cannot nest a SELECT within an INSERT
(in other words, you cannot insert the data returned by a SELECT).

The INSERT keyword is used to add new SQL records to a table.

Specify the names of the fields into which the data is to be inserted.
You cannot specify the values without the field name and expect the
server to insert the data into the correct fields by default.
INSERT INTO table_name ( column [ , column ]... )
VALUES (value [, value]... )

For example:
INSERT INTO emp_details ( first_name, last_name, dept,
salary)
VALUES ('David', 'Hughes', 'I.T.S.','12345')

Note – Single quotes are possible within a field item by escaping the
single quote: "\".

The number of values supplied must match the number of columns.

However, the column names are optional if every column value is
matched with an INSERT value.

B-6 Distributed Programming With Java Technology

The DELETE statement is used to remove records from a SQL table. The
syntax for the mSQL DELETE clause is
DELETE FROM table_name
WHERE column OPERATOR value
[ AND | OR column OPERATOR value ]...

Where

● OPERATOR can be <, >, =, <=, >=, <>, or the keyword LIKE.

For example:
DELETE FROM emp_details WHERE emp_id = '12345'

SQL Syntax B-7

The UPDATE statement is the SQL mechanism for changing the contents
of a SQL record. To change a particular record, you must identify what
record from the table you want to change. The mSQL UPDATE
statement cannot use a column name as a value. Only literal values
can by used as an UPDATE value. The syntax is
UPDATE table_name SET column=value [ , column=value ]...
WHERE column OPERATOR value
[ AND | OR column OPERATOR value ]...

Where

● OPERATOR can be <, >, =, <=, >=, <>, or the keyword LIKE.

For example:
UPDATE emp_details SET salary=30000 WHERE emp_id = '1234'

B-8 Distributed Programming With Java Technology

address
A location on a computer network, on a peripheral device, or in
computer memory.
address space
The range of memory locations to which a CPU can refer;
effectively, the amount of memory a CPU could use if all of the
memory were available.
applet
A Java program that can be included in an HTML page using
the applet tag.
application
Any specific use of the computer. The term is often used
synonymously with program.
application programmer’s interface (API)
The interface to a library or package of language-specific
functions or methods.
browser
A program used to view World Wide Web materials that is
capable of interpreting URLs and understanding different
Internet protocols.

Glossary-1
Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
class loader
A class loader is the foundation of the Java virtual machine
(JVM). A class loader is an executable class object that converts
a named class into the bits that make up an implementation of
that class. Class loaders enable the JVM to load a class without
having to know anything about the underlying file system
semantics. Class loaders extend the abstract class
java.lang.ClassLoader and implement the loadClass
method at minimum.
client
A software program that requests information or services from
another software application (server). For example, a browser is
a client that accesses data from HTTP servers.
Common Object Request Broker Architecture (CORBA)
The architecture and specifications aimed at software
developers and designers who want to produce applications
that comply with OMG standards for the ORB.
content handler
A specialized Java program that enhances Java technology
functionality by providing a means to understand a new
content type; for example, email, video, or audio files of
nonsupported types. It is responsible for reading data from a
stream (provided by a protocol handler) and returning an object
representation of the stream’s content.
distributed application
A program that makes calls to other address spaces, possibly on
another physical machine.
distributed computing
The technique of allowing applications running on one machine
to access applications that are running on another machine.
That is, client programs make calls to programs in other address
spaces, either local or remote.
distributed object computing
An extension of distributed computing, where objects are
implemented in an address space separate from the client
program.

Glossary-2 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
externalizable
Java objects are externalizable when they implement the
java.io.Externalizable interface and implement the
methods writeExternal(ObjectOutput out) and
readExternal(ObjectInput in). Externalizable objects are
serializable, but only the identity of the class is saved by
default. It is the responsibility of the class to save and restore
the contents of the object.
firewall
A machine or machines that run filtering and logging software,
which restrict and/or monitor traffic passing from one network
to another. A firewall is the single point through which all
traffic between two networks must pass, so that network and
application security policies can be implemented.
FTP
File Transfer Protocol. It is an Internet client-server protocol for
transferring files between computers.
GUI
Graphical user interface.
HTTP
Hypertext Transfer Protocol. The most common protocol used
on the World Wide Web to transfer hypertext documents.
idlgen
A command used to compile JavaIDL.
Interface Definition Language (IDL)
The Object Management Group (OMG) defined a set of
language constructs that can be used to define the interface
between an implementation of the interface and an application
that executes operations on the interface. The Interface
Definition Language (IDL) is the result of their work. The IDL
defines a “contract” of services. Each service encapsulates
operations and attributes and can specify exceptions that occur.
IDL is not a programmatic language; the IDL file must be
compiled to the specific language implementation desired.
Internet
The worldwide network of computers communicating using the
TCP/IP protocols.

Glossary Glossary-3
Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
Java
An object-oriented programming language developed by Sun
Microsystems to solve a number of problems in modern
programming practice.
JDBC API
A set of interfaces designed to insulate a database application
developer from a specific database vendor.
JavaIDL API
A set of classes and interfaces that enables developers to define
a set of remote interfaces using the CORBA IDL standard, and
maps IDL constructs onto a set of stubs and skeletons that can
be used to create a CORBA implementation.
multicast
A special form of broadcast where copies of the packet are
delivered to only a subset of all possible destinations.
network
A group of connected computers.
NFS
A distributed application that enables remote file systems to be
accessed by the end-user in the same way that a user would
access a local file system.
Object Management Group (OMG)
A nonprofit international consortium dedicated to promoting
the theory and practice of object technology for the
development of distributed computing systems.
Object Request Broker (ORB)
A program that provides the communications infrastructure
that enables objects to transparently make and receive requests
and responses in a distributed environment.
Object Serialization API
A set of classes and interfaces that enables developers to write
Java code that creates persistent storage for Java objects.
one-tier database design
A database written as a single unit, with both the database
engine and the user interface tightly coupled.
protocol
An agreed convention for inter-computer communication.

Glossary-4 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
protocol handler
A specialized Java program that enhances Java technology
functionality by providing a means to understand a new
protocol type; for example, IPX/SPX or Asynchronous Transfer
Mode (ATM). It is responsible for establishing a connection and
defining data stream syntax (how data will pass from one
endpoint to another). It returns a connection object that can then
be used for opening data streams.
RMI API
A set of classes and interfaces designed to enable developers to
make calls to remote objects that exist in the runtime of a
different virtual machine invocation.
Remote Procedure Call (RPC)
A paradigm for implementing the client-server model of
distributed computing. A request is sent to a remote system to
execute a designated procedure, using supplied arguments, and
return the result to the caller.
replicated
An object that has one or more exact copies of itself available
from multiple address spaces. For example, under NIS+, replica
servers provide the same information to clients as the master
server, and it is irrelevant to the client which server the
information comes from.
rmic
The RMI compiler that generates the RMI sub and skeleton
classes.
rmiregistry
An application that provides a simple naming lookup service of
remote objects in the RMI API.
Security Manager API
A class that enables developers to set and control the security
policy that must be followed by Java programs running on a
system.

Glossary Glossary-5
Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
serializable
Java objects are serializable if they implement the
java.io.Serializable interface and do not contain references
to nonserializable objects (for example, java.lang.Thread,
java.io.FileOutputStream) unless these references are
marked with the transient keyword. An object that is
serializable may be serialized.
serializing or serialized
Java objects that are serializable can be sent as a stream of bytes
over any type of java.io.Output stream and retrieved using
any type of java.io.InputStream. This includes files, pipes,
and sockets. An object is said to be serialized when it is
converted from its internal in-memory format to the stream of
bytes that can be sent over the wire.
server
In the client-server model for file systems, the server is a
machine with compute resources (and is sometimes called the
compute server) and large memory capacity.
skeleton
A server-side entity that contains a method that dispatches calls
to the actual remote object implementation.
socket
A software endpoint for network communication. Two
programs on different machines each open a socket in order to
communicate over the network. This is the low-level
mechanism that supports most networking programs.
stub
A proxy for a remote object that is responsible for forwarding
method invocations on remote objects to the server where the
actual remote object implementation resides.
TCP
Transport Control Protocol. A virtual circuit protocol of the
Internet protocol family. It provides reliable, flow-controlled, in-
order, two-way transmission of data in a byte stream.
thread
In programming, a process that is part of a larger process or
program.

Glossary-6 Distributed Programming With Java Technology

Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
three-tier database design
A database design that introduces an intermediary tier between
the database front end and the database engine. This
intermediary tier can support functionality such as mirroring,
secure transactions, and caching.
two-tier database design
A database design that separates the database front end from
the database engine, enabling the data to reside locally or
remotely.
UDP
User Data Protocol. A transport protocol in the Internet suite of
protocols using datagrams.

Glossary Glossary-7
Copyright 2000 Sun Microsystems, Inc. All Rights Reserved. Enterprise Services, March 2000, Revision C.1
Copyright 2000 Sun Microsystems Inc., 901 San Antonio Road, Palo Alto, California 94303, Etats-Unis. Tous droits
réservés.
Ce produit ou document est protégé par un copyright et distribué avec des licences qui en restreignent l’utilisation, la
copie, la distribution, et la décompilation. Aucune partie de ce produit ou document ne peut être reproduite sous aucune
forme, par quelque moyen que ce soit, sans l’autorisation préalable et écrite de Sun et de ses bailleurs de licence, s’il y en a.
Le logiciel détenu par des tiers, et qui comprend la technologie relative aux polices de caractères, est protégé par un
copyright et licencié par des fournisseurs de Sun.
Sun, Sun Microsystems, le logo Sun, Java, Java 2, JDK, Solaris, JavaSoft, JDBC, JavaIDL, JavaBeans, Write Once Run
Anywhere, 100% Pure Java, et Java Web Server sont des marques de fabrique ou des marques déposées de Sun
Microsystems, Inc. aux Etats-Unis et dans d’autres pays.
Toutes les marques SPARC sont utilisées sous licence sont des marques de fabrique ou des marques déposées de SPARC
International, Inc. aux Etats-Unis et dans d’autres pays.
Les produits portant les marques SPARC sont basés sur une architecture développée par Sun Microsystems, Inc.
Netscape Navigator est une marque de Netscape Communications Corporation.
UNIX est une marques déposée aux Etats-Unis et dans d’autres pays et licenciée exclusivement par X/Open Company,
Ltd.
L’interfaces d’utilisation graphique OPEN LOOK et Sun™ a été développée par Sun Microsystems, Inc. pour ses
utilisateurs et licenciés. Sun reconnaît les efforts de pionniers de Xerox pour larecherche et le développement du concept
des interfaces d’utilisation visuelle ou graphique pour l’industrie de l’informatique. Sun détient une licence non exclusive
de Xerox sur l’interface d’utilisation graphique Xerox, cette licence couvrant également les licenciés de Sun qui mettent
en place l’interface d’utilisation graphique OPEN LOOK et qui en outre se conforment aux licences écrites de Sun.
L’accord du gouvernement américain est requis avant l’exportation du produit.
Le système X Window est un produit de X Consortium, Inc.
LA DOCUMENTATION EST FOURNIE “EN L’ETAT” ET TOUTES AUTRES CONDITIONS, DECLARATIONS ET
GARANTIES EXPRESSES OU TACITES SONT FORMELLEMENT EXCLUES, DANS LA MESURE AUTORISEE PAR LA
LOI APPLICABLE, Y COMPRIS NOTAMMENT TOUTE GARANTIE IMPLICITE RELATIVE A LA QUALITE
MARCHANDE, A L’APTITUDE A UNE UTILISATION PARTICULIERE OU A L’ABSENCE DE CONTREFAÇON.

Please
Recycle

Murach J. Murach's MySQL. Training & Reference 4ed 2023
No ratings yet
Murach J. Murach's MySQL. Training & Reference 4ed 2023
645 pages
GM 3660 - Form - COMMODITY VALIDATION - CG4816Aug2018
100% (1)
GM 3660 - Form - COMMODITY VALIDATION - CG4816Aug2018
7 pages
Red Hat Jboss Enterprise Application Platform 7.1: Getting Started Guide
No ratings yet
Red Hat Jboss Enterprise Application Platform 7.1: Getting Started Guide
61 pages
Windows 10
No ratings yet
Windows 10
2,148 pages
Java
75% (4)
Java
239 pages
Web Services With IntelliJ IDEA
No ratings yet
Web Services With IntelliJ IDEA
22 pages
Advanced Queuing
No ratings yet
Advanced Queuing
1,210 pages
Oracle Service Bus 11g New Features For The Integration Developers
No ratings yet
Oracle Service Bus 11g New Features For The Integration Developers
70 pages
JSF Exercises v2 2
No ratings yet
JSF Exercises v2 2
43 pages
J2EE Web Component Developer
No ratings yet
J2EE Web Component Developer
302 pages
Java Programming Fundamentals
No ratings yet
Java Programming Fundamentals
21 pages
Web Development Using JSP
No ratings yet
Web Development Using JSP
800 pages
Java y UML
No ratings yet
Java y UML
155 pages
NetBeans IDE Java Quick Start Tutorial
No ratings yet
NetBeans IDE Java Quick Start Tutorial
6 pages
ChangeMan DS User Guide
No ratings yet
ChangeMan DS User Guide
496 pages
Introduction To Java Server Faces JSF Sun
No ratings yet
Introduction To Java Server Faces JSF Sun
36 pages
JDBC JSP
100% (3)
JDBC JSP
32 pages
NW - Js Essentials - Sample Chapter
No ratings yet
NW - Js Essentials - Sample Chapter
18 pages
Java Notes
100% (1)
Java Notes
20 pages
Rest PDF
No ratings yet
Rest PDF
17 pages
Hbase PDF
No ratings yet
Hbase PDF
33 pages
NHibernate Succinctly
No ratings yet
NHibernate Succinctly
182 pages
Unit-7: Java Web Frameworks: Spring MVC
No ratings yet
Unit-7: Java Web Frameworks: Spring MVC
26 pages
Java J2ee Syllabus JBK PDF
No ratings yet
Java J2ee Syllabus JBK PDF
3 pages
Java Developer
No ratings yet
Java Developer
7 pages
Log4j 2 Users Guide
100% (1)
Log4j 2 Users Guide
161 pages
Developing WebService Using RAD 7.5 (2.1) - Done
No ratings yet
Developing WebService Using RAD 7.5 (2.1) - Done
38 pages
Jasper Reports
No ratings yet
Jasper Reports
29 pages
Java Security
No ratings yet
Java Security
310 pages
Red Hat JBoss Enterprise Application Platform-7.1-How To Configure Server Security-en-US
No ratings yet
Red Hat JBoss Enterprise Application Platform-7.1-How To Configure Server Security-en-US
204 pages
Apex Overview
No ratings yet
Apex Overview
32 pages
UNIX and Shell Programming (Zer07)
No ratings yet
UNIX and Shell Programming (Zer07)
734 pages
React Native Training
No ratings yet
React Native Training
32 pages
Object-Oriented Programming in Java - InTL
No ratings yet
Object-Oriented Programming in Java - InTL
524 pages
Red Hat JBoss Enterprise Application Platform-7.1-Configuration Guide-en-US
No ratings yet
Red Hat JBoss Enterprise Application Platform-7.1-Configuration Guide-en-US
487 pages
Hartjes C. - The Grumpy Programmer's PHPUnit Cookbook - 2013
No ratings yet
Hartjes C. - The Grumpy Programmer's PHPUnit Cookbook - 2013
100 pages
The Java VM Architecture & Apis
No ratings yet
The Java VM Architecture & Apis
44 pages
Rest Servers in Delphi Xe Using Datasnap
No ratings yet
Rest Servers in Delphi Xe Using Datasnap
66 pages
Introduction
No ratings yet
Introduction
6 pages
D2kwordformat 140503090957 Phpapp01
No ratings yet
D2kwordformat 140503090957 Phpapp01
129 pages
Java WebServices Tutorial From JavaTpoint
No ratings yet
Java WebServices Tutorial From JavaTpoint
26 pages
Servlets Material
No ratings yet
Servlets Material
98 pages
Java Servlets Quick Notes
No ratings yet
Java Servlets Quick Notes
7 pages
Module 1 Introduction To Java Technology
No ratings yet
Module 1 Introduction To Java Technology
62 pages
Programming With Java - 2022.23 - CSE
No ratings yet
Programming With Java - 2022.23 - CSE
66 pages
Spring Boot 3.0 Crash Course
From Everand
Spring Boot 3.0 Crash Course
Kit Harrington
No ratings yet
Instant Razor View Engine How-to
From Everand
Instant Razor View Engine How-to
Abhimanyu Kumar Vatsa
No ratings yet
Apache Karaf Cookbook
From Everand
Apache Karaf Cookbook
Jamie Goodyear
No ratings yet
Mastering IBM WebSphere Portal: Expert Guidance to Build and Deploy Portal Applications
From Everand
Mastering IBM WebSphere Portal: Expert Guidance to Build and Deploy Portal Applications
Ron Ben Natan
4/5 (1)
JSF 2.0 Cookbook: LITE
From Everand
JSF 2.0 Cookbook: LITE
Anghel Leonard
No ratings yet
Microsoft Azure Fundamentals Certification and Beyond: A complete AZ-900 exam guide with online mock exams, flashcards, and hands-on activities
From Everand
Microsoft Azure Fundamentals Certification and Beyond: A complete AZ-900 exam guide with online mock exams, flashcards, and hands-on activities
Steve Miles
No ratings yet
CentOS Stream 9 Essentials: Learn to Install, Administer, and Deploy CentOS Stream 9 Systems
From Everand
CentOS Stream 9 Essentials: Learn to Install, Administer, and Deploy CentOS Stream 9 Systems
Neil Smyth
No ratings yet
Mastering Vue.js: Building Modern Web Applications
From Everand
Mastering Vue.js: Building Modern Web Applications
Pedro Martins
No ratings yet
PostgreSQL 9 High Availability Cookbook
From Everand
PostgreSQL 9 High Availability Cookbook
Shaun M. Thomas
5/5 (2)
Alfresco 3 Cookbook
From Everand
Alfresco 3 Cookbook
Snig Bhaumik
No ratings yet
OpenNebula 3 Cloud Computing
From Everand
OpenNebula 3 Cloud Computing
Giovanni Toraldo
No ratings yet
NW.js Essentials
From Everand
NW.js Essentials
Alessandro Benoit
No ratings yet
Java servlet Second Edition
From Everand
Java servlet Second Edition
Gerardus Blokdyk
No ratings yet
OpenStack Networking Cookbook
From Everand
OpenStack Networking Cookbook
Subramanian Sriram
No ratings yet
XSL Primer
From Everand
XSL Primer
Stephen Cote
No ratings yet
SOA-Based Enterprise Integration: A Step-by-Step Guide to Services-based Application
From Everand
SOA-Based Enterprise Integration: A Step-by-Step Guide to Services-based Application
Waseem Roshen
No ratings yet
Mastering GraphQL: From Basics to Expert Proficiency
From Everand
Mastering GraphQL: From Basics to Expert Proficiency
William Smith
No ratings yet
Warren Buffet's Lifestyle
No ratings yet
Warren Buffet's Lifestyle
11 pages
Subnetting and Supernetting
100% (5)
Subnetting and Supernetting
7 pages
J Apant LTR
No ratings yet
J Apant LTR
32 pages
Tree List Recursion
No ratings yet
Tree List Recursion
11 pages
Learning Flash
No ratings yet
Learning Flash
124 pages
Data Structures in C
94% (16)
Data Structures in C
47 pages
Learning Flash
No ratings yet
Learning Flash
124 pages
Oops
No ratings yet
Oops
3 pages
Nitish Steps To Install Hadoop
No ratings yet
Nitish Steps To Install Hadoop
3 pages
IXON TR001 - Remote Access Advanced Training v0.1
No ratings yet
IXON TR001 - Remote Access Advanced Training v0.1
58 pages
Siebel Testing - Best Practice: Siebel Has Provided Its Base Functionality (Vanilla Functionality) - For Siebel Testing
No ratings yet
Siebel Testing - Best Practice: Siebel Has Provided Its Base Functionality (Vanilla Functionality) - For Siebel Testing
3 pages
RAP As A Service For SQL Server: Prerequisites
No ratings yet
RAP As A Service For SQL Server: Prerequisites
10 pages
Resume I PDF
No ratings yet
Resume I PDF
2 pages
Delphi - Returning Data With Stored Procedures
No ratings yet
Delphi - Returning Data With Stored Procedures
16 pages
7. Sending Mail via GMail API
No ratings yet
7. Sending Mail via GMail API
8 pages
Learn SQL Server Administration in A Month of Lunches
No ratings yet
Learn SQL Server Administration in A Month of Lunches
1 page
Om All Joins
No ratings yet
Om All Joins
20 pages
xv6 Source Code
No ratings yet
xv6 Source Code
91 pages
Dev Ops Help
No ratings yet
Dev Ops Help
387 pages
Error Code 0xc000000f in Windows 7 !
No ratings yet
Error Code 0xc000000f in Windows 7 !
8 pages
Chapter Eleven: Handling Events
No ratings yet
Chapter Eleven: Handling Events
25 pages
Ossec Docs
No ratings yet
Ossec Docs
203 pages
Types of Computers
No ratings yet
Types of Computers
6 pages
9851 6637 01a RCS Trainer 2.2 - User Guide
No ratings yet
9851 6637 01a RCS Trainer 2.2 - User Guide
13 pages
CYS 7132 - Tutorial - 7
No ratings yet
CYS 7132 - Tutorial - 7
33 pages
Dell EMC Unity - Shell Lockdown (Rbash) On Unity OE 4.5 and Above (User Correctable) - Dell US
No ratings yet
Dell EMC Unity - Shell Lockdown (Rbash) On Unity OE 4.5 and Above (User Correctable) - Dell US
4 pages
106killer Game Programming in Java
No ratings yet
106killer Game Programming in Java
65 pages
PVI 14-09A - GT-I9301i Boot Recovery
No ratings yet
PVI 14-09A - GT-I9301i Boot Recovery
4 pages
Tinkercad - From Mind To Design in Minutes
No ratings yet
Tinkercad - From Mind To Design in Minutes
3 pages
An 138 Vinculum-II Debug Interface Description
No ratings yet
An 138 Vinculum-II Debug Interface Description
17 pages
Professional Windows Desktop and Server Hardening
No ratings yet
Professional Windows Desktop and Server Hardening
917 pages
The Software Method For The Nanocom Evolution mk2 Is Done in 2 Easy Steps
No ratings yet
The Software Method For The Nanocom Evolution mk2 Is Done in 2 Easy Steps
6 pages
Design of High-Confidence Embedded Operating System Based On Artificial Intelligence and Smart Chips
No ratings yet
Design of High-Confidence Embedded Operating System Based On Artificial Intelligence and Smart Chips
5 pages
Azure Fundamentals Notes: Principles of Cloud Computing
No ratings yet
Azure Fundamentals Notes: Principles of Cloud Computing
6 pages
Fbn
No ratings yet
Fbn
5 pages
Website Testing: Tech Mahindra Limited Confidential © Tech Mahindra Limited 2007
No ratings yet
Website Testing: Tech Mahindra Limited Confidential © Tech Mahindra Limited 2007
20 pages