ARYA GROUP OF COLLEGES

UNIT V
Distributed System
By
rashi Jain
Assistant Professor
Computer Science ENGG.
Department
Acerc, Jaipur
RTU syllabus Of UNIT V
Distributed Agreement: Concept of
Faults, failure and recovery, Byzantine
Faults, Adversaries, Byzantine
Agreement, Impossibility of Consensus
and Randomized Distributed Agreement
Replicated Data Management: concepts
and issues, Database Techniques,
Atomic multicast and Update
Propagation.
CORBA case study: Introduction,
Architecture, CORBA RMI,CORBA
Services
Lecture Plan of Unit V
 Concepts of faults ,failure and recovery
 Fault Classification
Lecture 1
 failure models in distributed system
 Recovery in Distributed system
 Byzantine faults
 Adversaries
 Byzantine Agreement Lecture
2
 Impossibility of Consensus
 Randomized Distributed Agreement
 Exponential time shared memory consensus
Lecture 3
 Polynomial time shared memory consensus
Lecture Plan of Unit V(Cont..)
Memory management Algorithm
Replicated data management
Lecture 4
Concepts and Issues
Database Techniques
Atomic multicast
Update Propagation
Epidemic Algorithm Lecture 5
Antientropy
Update Logs
Case study : CORBA Lecture 6
Concept of Fault, Failure & Recovery

Definition:
◦ Fault tolerance is the ability of a
system to perform its function
correctly even in the presence of
internal faults.
The concept of fault, failure &
Recovery is described by using
following points.
◦ Fault Classification
◦ Failure model in distributed system
◦ Recovery in Distributed system.
Fault Classification
Fault can be classified as transient or
permanent.
A transient fault will eventually
disappear without any apparent
intervention.
A permanent one will remain unless it
is removed by some external agency.
Classification of faults
◦ Crash Fault
◦ Omission fault
◦ Timing Fault
◦ Byzantine fault
Failure Model in Distributed System
Definition
◦ It define a precise specification of the
faults that can be exhibited by
processes and communication
channels.
◦ Omission Failure
◦ Timing Failure
◦ Arbitrary Failure
Recovery In Distributed System
Definition
◦ A transaction processing system will go
through a recovery of the database to cope
when the system fails, it involves the backup,
journals, checkpoint and recovery manager.
Types of Recovery Techniques:
◦ Journal: A journal maintain an audit trail of
transaction and database changes.
◦ Checkpoint: It is to provide a snapshot of the
data within the database.
◦ Recovery Manager: A recovery manager is a
program which restores the database to a
correct condition which can restart the
transaction processing.
Byzantine Faults
Definition:
◦ A Byzantine fault is an arbitrary fault
that occur during execution of an
algorithm by a distributed system.
It encompasses both omission
failures and commission failure.
When a byzantine fault has
occurred the system may respond
in any unpredictable way, unless it
is designed to have byzantine
fault tolerance.
Adversaries
Definition:
◦ An adversary is a demon that has a great
deal of control over the environment in
which your protocol executes and will try
to make your protocol fail.
The adversary has the power to
examine the global state and to
schedule the execution of the protocol.
Adversary is the ability to destroy or
modify message and change the
protocol that some of the processors
execute.
Byzantine Agreement
Definition:
◦ It refers to distributed agreement
algorithm in which a failed processor
can send arbitrary message into the
network.
◦ A failed processor can try to prevent
the nonfailed processor from
reaching an agreement.
Impossibility Of Consensus

Consensus is the task of getting

all processes in a group to agree
on some specific value based on
the votes of each processes.
The impossibility result depend
completely on the fact that the
distributed system is
asynchronous. While processor
can fail the system is otherwise
reliable.
Randomized Distributed Agreement
A randomized algorithm is one
which contains an assignment to a
variable based on the outcome of
tossing a fair coin or a random
number generator.
The randomized distributed
agreement can be described by
using some point that is as follows.
◦ Exponential Time Shared memory
Consensus.
◦ Polynomial Time Shared memory
Consensus.
Randomized Distributed
Agreement(Cont..)
Exponential Time Shared memory
Consensus:
◦ It is simple-minded approach to writing
a consensus protocol in a shared
memory system.
◦ We initialize by choosing a value to
prefer, Vp we look at the value
preferred by every other processors
{Vj}. If all other processors have the
same value Vi =Vp, we decide on value
Vp. Otherwise we flip a coin and set Vp
to the value of the coin Flip.
Polynomial Time Shared memory
Consensus
This protocol make it that every
processor obtains the same value on its
flip() on a given round r.
If every processor obtain the same value
for flip() on an arbitrary round r will
probability at least Psame, the
SM_Consensus protocol will require
O(1/Psame) rounds.
If Psame is independent of N, only an
expected constant number of round are
required. So it is worthwhile to spend a
fair amount of effort to get every
processor to flip the same coin.
Message Passing Randomized Consensus
Itis similar to the shared memory
consensus protocol except shared
variable is the V array.
The element V[i] is written to by
processor i only and is read by
every other processor. So array V
makes shared memory look like a
broadcast medium.
A naive translation of the
SM_consensus protocol to the
MP_consensus protocol is o
broadcast all updates.
Memory Management Algorithm
Definition
◦ In multiprogramming system, to
share the processor with a number of
processes that must be kept in
memory. These memory
management is achieved through
memory management algorithms.
Methodsof Memory
Management:
◦ Fixed Partitioning
◦ Variable Partitioning
◦ Paging
Memory Management
Algorithm(Cont..)
Fixed Partitioning
◦ In this method, memory is divided
into partitions whose sizes are fixed.
Operating system is placed into the
lowest bytes of the memory.
Processes are classified on entry of
the system according to their
memory they requirements.
Operating System

Small
(nkB) Medium
(3nkB) Large

(6nkB)
Memory Management
Algorithm(Cont..)
Variable Partitioning
◦ In it, the memory size may vary
dynamically.
◦ In it, we keep a table(linked list)
indicating used/free areas in
memory.
There are three algorithms for
searching the list of free blocks
for a specific amount of memory.
◦ First Fit: Allocate the first free block
that is large enough for the new
process. This is the fast algorithm.
◦ Best Fit
 Memory Management
Algorithm(Cont..)
Paging
◦ Paging permit a program to allocate
noncontiguous blocks of memory. The
OS divide the program into pages
which are blocks of small and fixed
size. Then it divide the physical
memory into frames which are block of
size equals to page size.
The OS uses a page table to map
program pages to memory frames.
Replicated Data Management
Definition
◦ Replication is the process of copying &
maintaining database objects in multiple
database that make up a distributed
database system.
◦ Replication can improve the performance &
protect the availability of application because
alternate data access option exist.
The replicated data management is
described by using some point.
◦ Concept & Issues
◦ Database Techniques
◦ Atomic Multicast
◦ Update Propagation
Replicated Data Management(Cont..)
Concept & Issues:
◦ Basic replication environment support
application that require read-only access to
the table data that originates from a primary
site.
The concept of basic replication
environments
◦ Use of basic replication
 Information distribution, off loading & Information
transport.
◦ Read-only table snapshots
 It is a local copy of table data that originates from
one or more remote master tables.
◦ Snapshots Refreshes
 A table snapshot is a transaction- consistent
reflection of its master data as that data existed at a
specific point in time.
Replicated Data Management(Cont..)
Database Techniques:
◦ It is a technique that are used for
managing replicated data, that is called
database techniques.
Itwill examine the fundamentals
algorithms for maintaining data
consistency in a distributed database.
Types of Database techniques:
◦ Database logging & Recovery
◦ Two-phase Commit
◦ Three-phase Commit
◦ Replicated-Data- Management
◦ Dynamic Quorum Changes
Atomic Multicast
Definition:
◦ If one processor on the delivery list
receives a reliable multicast messages,
every processor on the delivery list
receives the message.
Some point to describe the Atomic
multicast.
◦ Virtual Synchrony
◦ Ordered Multicasts
◦ Reliable & Casual multicast
◦ Group Membership
◦ Atomic group multicasts.
Update Propagation
The use of update propagation is
to replicate the database it
manages.
Updates need to be propagated
to all sites.
 The update propagation can be
described by using some point
that is as follows.
◦ Epidemic Algorithms
◦ Antientropy
◦ Update Logs
Update Propagation(Cont..)
Epidemic Algorithms
◦ When a server updates a data items d, it
should start the process of distributing the
update.
Antientropy:
◦ A distributing updates is to have one site
contact another to exchange recent
updates, that is called Antientropy.
Update Logs
◦ Every processor p keeps a log L of the
updates that it has processed. The log is
an listing of event records, where each
event corresponding to an update.
Case Study: CORBA
Definition
◦ CORBA is a middleware design that
allows application programs to
communicate with one another
irrespective of their programming
language, their hardware and
software platforms, the networks
they communicate over and their
implementers.
◦ Applications are built from CORBA
objects, which implement interfaces
defined in CORBA’s interface
Case Study: CORBA(Cont..)
The Common object Request
Broker architecture is described
as follows:
◦ Architecture
◦ CORBA RMI
◦ CORBA Services
Case Study: CORBA(Cont..)
Architecture of CORBA:
◦ It designed to support the role of an
object request broker that enables
clients to invoke methods in remote
objects, where both clients and
server can be implemented in a
variety of programming language.
Case Study: CORBA(Cont..)
In the CORBA architecture contains
three additional components:
◦ The object adapter
◦ The implementation Repository
◦ The interface Repository
CORBA provides for both static &
Dynamic invocations.
◦ Static invocation are used when the
remote interface of the CORBA object is
known at compile time, enabling client
stub & Server skeleton to be used.
◦ If the remote interface is not known at
compile time, dynamic invocation must
be used.
Case Study: CORBA(Cont..)
Object Adapter:
◦ The role of an object adapter is to
bridge the gap between CORBA objects
with IDL interfaces and the
programming language interface of the
corresponding servant classes.
Skeleton:
◦ Skeleton classes are generated in the
language of the server by an IDL
compiler. The skeleton unmarshals the
arguments in request messages and
marshals exception and result in reply
messages.
Case Study: CORBA(Cont..)
Client Stub:
◦ In the client language, a set of stub
procedures is generated from an IDL
interfaces by an IDL compiler for the client
language.
Implementation Repository:
◦ It is responsible for activating registered
servers on demand and for locating servers
that are currently running.
◦ The object adapter name is used to refer to
servers when registering and activating
them.
Interface Repository
◦ It is to provide information about registered
IDL interfaces to client and servers that
Case Study: CORBA(Cont..)
CORBA RMI:
◦ CORBA RMI requires more of the
programmer than programming in a
single language RMI system such as
Java RMI.
CORBA Services:
◦ CORBA naming Services
◦ CORBA Event Services
◦ CORBA notification Services
◦ CORBA Security Services.
Thank You