ZSeries and ZOS Reference Guide

Download as pdf or txt
Download as pdf or txt
You are on page 1of 49

Enterprise Servers

© Copyright IBM Corporation 2001


IBM Corporation
Integrated Marketing Communications,
Server Group
Route 100
IBM
Somers, NY 10589
Produced in the United States of America, 10-01
All Rights Reserved
zSeries 900 and z/OS
References in this publication to IBM products or services do
not imply that IBM intends to make them available in every
country in which IBM operates. Consult your local IBM busi-
Reference Guide
ness contact for information on the products, features, and
services available in your area. October 2001
IBM, the IBM logo, the e-business logo, AnyNet, APPN, CICS, DB2,
DB2 Connect, DB2 Universal Database, DFSMSdfp, DFSMSdss,
DFSMShsm, DFSMSrmm, DFSORT, DFSMS/MVS, ENCINA, Enter-
prise Storage Server, ESCON, FlashCopy, FFST, FICON, GDDM,
GDPS, Geographically Dispersed Parallel Sysplex, HiperSockets,
Hiperspace, IMS, InfoPrint, Intelligent Miner, Language Environ-
ment, Magstar, MQSeries, Multiprise, MVS, Netfinity, NetSpool,
NetView, OS/2, OS/390, Parallel Sysplex, PrintWay, PR/SM, Proces-
sor Resource/Systems Manager, pSeries, RACF, RMF,
RS/6000, S/390, S/390 Parallel Enterprise Server, SecureWay,
Sysplex Timer, Systempac, Tivoli, Virtual Image Facility, VM/ESA,
VSE/ESA, VTAM, WebSphere, xSeries, z/Architecture, z/OS,
z/VM, and zSeries are trademarks or registered trademarks of
the International Business Machines Corporation.
Java and all Java-based trademarks and logos are trademarks
or registered trademarks of Sun Microsystems, Inc. in the
United States or other countries.
Tux appears courtesy of Larry Eung.
UNIX is a registered trademark of The Open Group.
Windows NT is a registered trademarks of Microsoft Corporation.
Linux is a registered trademark of Linus Torvalds.
Other trademarks and registered trademarks are the properties
of their respective companies.
IBM hardware products are manufactured from new parts, or
new and used parts. Regardless, our warranty terms apply.
Photographs shown are engineering prototypes. Changes may
be incorporated in production models.
This equipment is subject to all applicable FCC rules and will
comply with them upon delivery.
Information concerning non-IBM products was obtained from
the suppliers of those products. Questions concerning those
products should be directed to those suppliers.
All customer examples described are presented as illustrations
of how these customers have used IBM products and the re-
sults they may have achieved. Actual environmental costs and
performance characteristics may vary by customer.
All statements regarding IBM's future direction and intent are
subject to change or withdrawal without notice,
and represent goals and objectives only.
Prices subject to change without notice. Contact your IBM
representative or Business Partner for the
most current pricing in your geography.

G326-3092-01
zSeries Overview one gigabit per second. Furthermore, the availability of
native FICON™ devices and FICON CTC can increase
The IBM® ^ zSeries™ 900 family of servers offers I/O performance, simplify/consolidate channel
26 air-cooled models, from one-way to 16-way, utilizing configuration, and help reduce total cost of ownership.
from three to twenty processors. These servers can be The total result is ultra high-speed communications
configured in numerous ways to offer unparalleled within the server, between servers, to devices, and out
flexibility to speed deployment of e-business solutions. to users, thereby allowing greater integration between
Each z900 can operate independently or as part of a traditional and Web applications to maximize
Parallel Sysplex® cluster of servers with as many as e-business effectiveness.
640 processors. In addition to supporting z/OS™, the
z900 can host hundreds of Linux® images running The z900 family of servers also automatically directs
open source applications using Linux or z/VM™. resources to priority work through Intelligent Resource
Director (IRD). The z900 IRD combines the strengths of
The z900 family is based on the z/Architecture™, which three key technologies: z/OS Workload Manager,
supports a new standard of performance and Logical Partitioning and Parallel Sysplex clustering.
integration by expanding on the balanced system
approach of the S/390® architecture. New Tools for Managing e-business
The IBM ^ product line is backed by a
The z900 can eliminate bottlenecks associated with
comprehensive suite of offerings and resources that
the lack of addressable memory through its virtually
provide value at every stage of IT implementation.
unlimited 64-bit addressing capability, providing plenty
These tools can help customers test possible solutions,
of “headroom” for unpredictable workloads and growing
obtain financing, plan and implement applications and
enterprise applications.
middleware, manage capacity and availability, improve
A robust I/O subsystem complements the greater performance and obtain technical support across the
number of processors and larger main memory. High- entire infrastructure. The result is an easier way to
speed interconnects for TCP/IP communication, known handle the complexities and rapid growth of
as HiperSockets™, let TCP/IP traffic travel between e-business. In addition, IBM Global Services experts
partitions at memory speed, rather than network speed. can help with business and IT consulting, business
A high performance Gigabit Ethernet feature is one of transformation and total systems management services,
the first in the industry capable of achieving line speed: as well as customized e-business solutions.

* The S/390 Parallel Enterprise Server and OS/390 Reference Guide


(G326-3070-12) describes all functions and features available on the prior
S/390 CMOS servers which are also available on the z900.

4 5
Ease of Use and Self-Management systems management. zSeries plays a major role in
To help organizations deal effectively with complexity, Project eLiza, since the self-management capabilities
IBM has announced Project eLiza, a blueprint for self- available for the zSeries will function as a model for
managing systems. The goal is to use technology to other IBM servers.
manage technology, creating an intelligent, self-
managing IT infrastructure that minimizes complexity The latest z900 and z/OS V1R2 announcement gives
and gives customers the ability to manage the ability to configure, connect, extend, operate and
environments that are hundreds of times more complex optimize the computing resources to efficiently meet
and more broadly distributed than exist today. This the “always-on” demands of e-business.
enables increased utilization of technology without the
spiraling pressure on critical skills, software and One of the key functions of z/OS is Intelligent Resource
service/support costs. Director (IRD), an exclusive IBM technology that makes
the z900 the only server capable of automatically
Project eLiza represents a major shift in the way the reallocating processing power to a given application
industry approaches reliability, availability and on the fly, based on the workload demands being
serviceability (RAS). It harnesses the strengths of IBM experienced by the system at that exact moment. This
and its partners to deliver open, standards-based advanced technology, often described as the "living,
servers and operating systems that are self-configuring, breathing server," allows the z900 and z/OS to provide
self-protecting, self-healing and self-optimizing. Project nearly unlimited capacity and nondisruptive scalability,
eLiza technology helps ensure that critical operations according to priorities determined by the customer.
continue without interruption and with minimal need for
operator intervention.

Project eLiza will help customers dramatically reduce


the cost and complexity of their e-business
infrastructures, and overcome the challenges of

6 7
z/Architecture

The zSeries is based on the z/Architecture, which is z/Architecture Operating System Support
designed to eliminate bottlenecks associated with the The z/Architecture is a tri-modal architecture capable
lack of addressable memory and automatically directs of executing in 24-bit, 31-bit, or 64-bit addressing
resources to priority work through Intelligent Resource modes. Operating systems and middleware products
Director (IRD). The z/Architecture is a 64-bit superset of have been modified to exploit the new capabilities of
ESA/390. the z/Architecture. Immediate benefit can be realized by
the elimination of the overhead of Central Storage to
This architecture has been implemented on the z900 Expanded Storage page movement and the relief
to allow full 64-bit real and virtual storage support. A provided for those constrained by the 2 GB real storage
maximum 64 GB of real storage is available on z900 limit of ESA/390. Application programs will run
servers. z900 can define any LPAR as having 31-bit or unmodified on the z900.
64-bit addressability.
Expanded Storage (ES) is still supported for operating
z/Architecture has: systems running in ESA/390 mode (31-bit). For
• 64-bit general registers. z/Architecture mode (64-bit), ES is supported by z/VM.
ES is not supported by z/OS in z/Architecture mode.
• New 64-bit integer instructions. Most ESA/390 architec-
ture instructions with 32-bit operands have new 64-bit Although z/OS does not support Expanded Storage
and 32- to 64-bit analogs. when running under the new architecture, all of the
• 64-bit addressing is supported for both operands and Hiperspace™ and VIO APIs, as well as the Move Page
instructions for both real addressing and virtual ad- (MVPG) instruction, continue to operate in a compatible
dressing. manner. There is no need to change products that use
• 64-bit address generation. z/Architecture provides 64-bit Hiperspaces.
virtual addressing in an address space, and 64-bit real
addressing. Some of the exploiters of z/Architecture for z/OS and
• 64-bit control registers. z/Architecture control registers OS/390® Release 10 include:
can specify regions, segments, or can force virtual • DB2® Universal Database™ Server for OS/390
addresses to be treated as real addresses. • IMS™
• The prefix area is expanded from 4K to 8K bytes. • Hierarchical File System (HFS)
• New instructions provide quad-word storage consistency. • Virtual Storage Access Method ( VSAM)
• The 64-bit I/O architecture allows CCW indirect data • Remote Dual Copy (XRC)
addressing to designate data addresses above 2GB • Tape and DASD access methods
for both format-0 and format-1 CCWs. Operating System
ESA/390 z/Arch.
(31-bit) (64-bit)
• IEEE Floating Point architecture adds twelve new
instructions for 64-bit integer conversion. z/OS Version 1 Release 1, 2 and 3 No Yes

• The 64-bit SIE architecture allows a z/Architecture OS/390 Version 2 Release 10 Yes Yes

server to support both ESA/390 (31-bit) and OS/390 Version 2 Release 6, 7, 8, and 9 Yes No
z/Architecture (64-bit) guests. Zone Relocation is Linux for zSeries No Yes
expanded to 64-bit for LPAR and VM/ESA®. Linux for S/390 Yes No
• Use of 64-bit operands and general registers for all
S/390 Virtual Image Facility for Linux Yes No
Cryptographic Coprocessors instructions and Periph-
z/VM Version 4 Release 1 and 2 Yes Yes
eral Component Interconnect Cryptographic
Coprocessors (PCICC) instructions is added. z/VM Version 3 Release 1 Yes Yes

VM/ESA Version 2 Release 3 and 4 Yes No


The implementation of 64-bit z/Architecture can eliminate VSE/ESA Version 2 Release 3, 4 and 5 Yes No
any bottlenecks associated with lack of addressable
TPF Version 4 Release 1 (ESA mode only) Yes No
memory by making the addressing capability virtually
* SOD z/OS 1.4 supports ESA/390 or G5/G6 and MP 3000
unlimited (16 Exabytes from the current capability of 2 GB).

8 9
Intelligent Resource Director zSeries IRD Scope
Exclusive to IBM's z/Architecture is Intelligent Resource
Director (IRD), a function that optimizes processor and LPAR Cluster
channel resource utilization across Logical Partitions
(LPARs) based on workload priorities. IRD combines z/OS
the strengths of the z900 LPARs, Parallel Sysplex
clustering, and z/OS Workload Manager.

Intelligent Resource Director uses the concept of an z/OS


LPAR cluster, the subset of z/OS systems in a Parallel
Sysplex cluster that are running as LPARs on the same
z900 server. In a Parallel Sysplex environment,
Linux
Workload Manager directs work to the appropriate
resources based on business policy. With IRD,
resources are directed to the priority work. Together,
Parallel Sysplex technology and IRD provide the OS/390
flexibility and responsiveness to e-business workloads
unrivaled in the industry.
ICF
IRD has three major functions: LPAR CPU
Management, Dynamic Channel Path Management,
and Channel Subsystem Priority Queuing.
load. For example, in an environment where an
LPAR CPU Management installation normally requires four channels to several
LPAR CPU Management allows WLM working in goal control units, but occasionally needs as many as six,
mode to manage the processor weighting and logical system programmers must currently define all six
processors across an LPAR cluster. With z/OS Version 1 channels to each control unit that may require them.
Release 2, WLM can even direct CPU resources With Dynamic Channel Path Management, the system
outside a z/OS LPAR cluster, to an LPAR running either programmer need only define the four channels to the
z/VM or Linux. CPU resources are automatically moved control units, and indicate that DCM may add an
toward LPARs with the greatest need by adjusting the additional two. As the control unit becomes more
partition's weight. WLM also manages the available heavily used, DCM may assign channels from a pool of
processors by adjusting the number of logical CPs in managed channels, identified by the system
each LPAR. This optimizes the processor speed and programmer, to the control unit. If the work shifts to other
multiprogramming level for each workload, reduces MP control units, DCM will unassign them from lesser
overhead, and gives z/OS more control over how CP utilized control units and assign them to what are now
resources are distributed to meet your business goals. the more heavily used ones. This helps reduce the
requirement for greater than 256 channels. DCM can
z/OS V1R2 enhances the LPAR CPU management also reduce the cost of the fibre infrastructure required
capabilities and will allow the dynamic assignment of for connectivity between multiple data centers.
CPU resources to non-z/OS partitions such as Linux.
Channel Subsystem Priority Queuing
Dynamic Channel Path Management The notion of I/O Priority Queuing is not new; it has
In the past, and on other architectures, I/O paths are been in place in OS/390 for many years. With IRD, this
defined with a fixed relationship between processors capability is extended into the I/O channel subsystem.
and devices. With z/OS and the z900, paths may be Now, when higher priority workloads are running in an
dynamically assigned to control units to reflect the I/O LPAR cluster, their I/Os will be given higher priority, and

10 11
will be sent to the attached I/O devices (normally disk Since HiperSockets does not use an external network,
but also tape and network devices) ahead of I/O for it can free up system and network resources,
lower priority workloads. LPAR priorities are managed eliminating attachment cost while improving availability
by WLM in goal mode. and performance. HiperSockets can have significant
value in server consolidation, for example, by
Channel Subsystem Priority Queuing provides two connecting LPARs multiple Linux virtual servers under
advantages. First, customers who did not share I/O z/VM to z/OS machines. Furthermore, HiperSockets will
connectivity via MIF (Multiple Image Facility) out of be utilized by TCP/IP in place of XCF for sysplex
concern that a lower priority I/O intensive workload connectivity between images which exist in the same
might preempt the I/O of higher priority workloads, can server, thus z/OS uses HiperSockets for connectivity
now share the channels and reduce costs. Second, between sysplex images in the same server and uses
high priority workloads may even benefit with improved XCF for connectivity between images in different
performance if there were I/O contention with lower servers. Management and administration cost
priority workloads. Initially, Channel Subsystem Priority reductions over existing configurations are possible.
Queuing is implemented for Parallel OEMI and ESCON®,
FICON and FICON Express channels.

Channel Subsystem Priority Queuing complements the


IBM Enterprise Storage Server™ capability to manage
I/O priority across CECs.

With IRD, the combination of z/OS and the z900


working in synergy extends the industry leading
workload management tradition of S/390 and OS/390 to
ensure that the most important work on a server meets
its goals, to increase the efficiency of existing
hardware, and to reduce the amount of intervention in a
constantly changing environment.

HiperSockets
HiperSockets, a new feature unique to the zSeries,
provides a “TCP/IP network in the server” that allows
high speed any-to-any connectivity among virtual
servers (TCP/IP images) within a z900 without any
physical cabling. HiperSockets minimizes network
latency and maximizes bandwidth between
combinations of Linux, z/OS and z/VM virtual servers.
These OS images can be first level (directly under an
LPAR), or second level images (under z/VM).

With up to four HiperSockets per LPAR connection, one


could separate traffic to different HiperSockets for
security (separation of LAN traffic, no external wire-
tapping, monitoring) and performance and
management reasons (separate sysplex traffic Linux or
non-sysplex LPAR traffic).

12 13
z900 Support for Linux

HiperSockets acts like any other TCP/IP network Linux for zSeries
interface, so TCP/IP features like IP Security (IPSec) in Linux and zSeries make a great team. The flexibility and
Virtual Private Networks (VPN) and SSL can be used to openness of Linux bring with it access to a very large
provide heightened security for flows within the same portfolio of applications. zSeries incorporates the
CHPID. HiperSockets supports multiple frame sizes, qualities of service that deliver an industrial strength
which is configured on a per HiperSockets CHPID environment for these Linux applications. In addition
basis. This support gives the user the flexibility to zSeries enables massive scalability within a single
optimize and tune each HiperSockets to the server. Hundreds of Linux images can run
predominate traffic profile, for example distinguish simultaneously, providing unique server consolidation
between “high bandwidth” workloads such as FTP capabilities and reducing both cost and complexity.
versus lower bandwidth interactive workloads.
Of course, no matter which Linux applications are
The HiperSockets function provides many possibilities brought to the zSeries platform, they all benefit from
for improved integration between workloads in different high speed access to the corporate data that typically
LPARs, bounded only by the combinations of operating resides on zSeries.
systems and their respective applications.
HiperSockets will provide the fastest z900 connection IBM has developed a series of patches that enable
between e-business and ERP solutions sharing Linux to run on the S/390 platform. These were released
information while running on the same server. to the Open Source community in December, 1999.
WebSphere® http and Web Application Servers or Linux for zSeries supports the 64-bit architecture
Apache http servers can be running in a Linux LPAR or available on zSeries processors. This architecture
z/VM guest machine and will be able to use eliminates the existing main storage limitation of 2 GB.
HiperSockets for very fast TCP/IP traffic transfer to a Linux for zSeries provides full exploitation of the
DB2 database server running in a z/OS V1.2 LPAR. architecture in both real and virtual modes. Linux for
System performance is optimized because this allows zSeries is based on the Linux 2.4 kernel. Linux for
you to keep your Web and transaction application S/390 is also able to execute on zSeries and S/390 in
environments in close proximity to your data and 32-bit mode:
eliminates any exposure to network related outages,
thus improving availability. IBM Software
• IBM WebSphere Application Server Advanced Edition 3.5
The new z/OS HiperSockets Accelerator function can • WebSphere Commerce Suite, Pro Edition for Linux for
improve performance and cost efficiencies when zSeries Version 5.1
attaching a high number of TCP/IP images via • WebSphere Personalization Version 3.5
HiperSockets to a “front end” z/OS system for shared • MQSeries® Client for Linux for S/390 Version 5.2
access to a set of OSA-Express adapters. • WebSphere Host-On-Demand Version 5.0.3
• CICS® Transaction Gateway Version 4.0
• Tivoli® Storage Manager Linux for zSeries and S/390
Client Version 4.2
• IMS Connect
• DB2 Universal Database Enterprise Edition Version 7.2
• DB2 Connect™ Unlimited Edition Version 2
• DB2 Connect Web Starter Kit Version 7.2
• DB2 Intelligent Miner™ Scoring Version 7.2
• DB2 Net Search Extender Version 7.2

14 15
Linux Distribution Partners HiperSockets
• SuSE Linux Enterprise Server for S/390 HiperSockets can be used for communication between
Product Information available at suse.de/en/ Linux image and Linux or z/OS images whether Linux is
produkte/susesoft/S390/. There are many different op- running in an IFL LPAR, natively or under z/VM.
tions for ordering. Customers should contact SuSE by
sending an e-mail to s390-sales@suse.de. Crypto
• TurboLinux Server 6 for zSeries and S/390 Linux for zSeries running on standard z900 engines is
Product Information at turbolinux.com/products/s390. capable of exploiting the hardware cryptographic
Customers should contact the TurboLinux sales depart- feature provided by the PCI card for SSL acceleration.
ment by sending an e-mail to sales@turbolinux.com. This enables customers implementing e-business
• Red Hat Server for zSeries applications on Linux for zSeries to utilize this
Product Information at redhat.com. enhanced security of the hardware.
z/VM Version 4
z/VM enables large scale horizontal growth of Linux Linux Support
images when using Linux for zSeries. Only z/VM gives Environment
the capability of running hundreds of Linux for zSeries • zSeries or S/390 single image
or Linux for S/390 images. This version of z/VM is priced • zSeries or S/390 LPAR
on a per-engine basis (onetime charge) and will • Virtual Image Facility image
support IBM Integrated Facility for Linux (IFL) processor • VM/ESA or z/VM guest
features for Linux-based workloads and standard Block devices
engines for all other zSeries and S/390 workloads. • VM minidisks
• ECKD 3380 or 3390 DASDs
IBM S/390 Integrated Facility for Linux
• VM virtual disk in storage
This optional feature provides a way to add processing
Network devices
capacity, exclusively for Linux workload, with no effect
• Virtual CTC
on the model designation. No traditional zSeries
• ESCON CTC
workload will be able to run in this area. Consequently,
• OSA-Express (Gigabit Ethernet, Ethernet, Fast Ethernet,
these engines will not affect the IBM S/390 and zSeries
Token-Ring)
software charges for workload running on the other
• HiperSockets
engines in the system.
• 3172
• IUCV
OSA-Express Gigabit Ethernet for Linux
• Character devices
Driver support is provided for the functions of the OSA-
• 3215 console
Express Gigabit Ethernet feature. This driver supports
• Integrated console
the IPv4 protocol, delivering the advantages of more
rapid communication across a network. This Additional information is available at ibm.com/linux/
improvement may be between virtual Linux instances and at ibm.com/zseries/linux/.
on a single machine (either in LPAR or virtual mode)
communicating across a network, or a Linux for
zSeries or Linux for S/390 instance communicating
with another physical system across a network.

16 17
Parallel Sysplex Cluster Technology

Parallel Sysplex clustering was designed to bring the cluster of multiple footprints as a single operational image.
power of parallel processing to business-critical The common time source also enables proper sequencing
zSeries and S/390 applications. A Parallel Sysplex and time stamping of updates to shared databases, a
cluster consists of up to 32 z/OS and/or OS/390 images feature critical to recoverability of the shared data.
coupled to one or more Coupling Facilities (CFs or
ICFs) using high speed specialized links for Coupling Facility Configuration Alternatives
communication. The Coupling Facilities, at the heart of IBM offers different options for configuring a functioning
the Parallel Sysplex cluster, enable high-speed, Coupling Facility:
read/write data sharing and resource sharing among • Standalone Coupling Facility: z900 Model 100, and 9672-
all the z/OS and OS/390 images in a cluster. All images R06 models provide a physically isolated, totally indepen-
are also connected to a Sysplex Timer® to ensure time dent CF environment. There are no software charges
synchronization. associated with standalone CF models. An ICF or CF par-
Parallel Sysplex Resource Sharing enables multiple tition sharing a server with z/OS or OS/390 images not in
system resources to be managed as a single logical the sysplex acts like a logical standalone CF.
resource shared among all of the images. Some • Internal Coupling Facility (ICF): Customers considering
examples of resource sharing include Automatic Tape clustering technology can get started with Parallel
Switching, GRS star, and Enhanced Catalog Sharing; all of Sysplex technology at a lower cost. An ICF feature is a
which provide simplified systems management, processor that can only run Parallel Sysplex coupling
increased performance and/or scalability. For more detail, code (CFCC) in a partition. Since CF LPARs on ICFs are
please see S/390 Value of Resource Sharing White Paper restricted to running only Parallel Sysplex coupling
– GF22-5115-00 on the Parallel Sysplex home page at code, there are no software charges associated with
ibm.com/servers/eservers/zseries/pso. ICFs. ICFs are ideal for Intelligent Resource Director
and resource sharing environments.
Although there is a significant value in a single
footprint and multi-footprint environment with resource • Coupling Facility partition on a z900 or 9672 server
sharing, those customers looking for high availability using standard LPAR: A CF can be configured to run in
must move on to a database data sharing configuration. either a dedicated or shared CP partition. Software
With the Parallel Sysplex environment, combined with charges apply. This may be a good alternative for test
the Workload Manager and CICS TS or IMS TM, configurations that require very little CF processing
incoming work can be dynamically routed to the z/OS resource or for providing hot-standby CF backup using
or the OS/390 image most capable of handling the the Dynamic Coupling Facility Dispatching function.
work. This dynamic workload balancing, along with the A Coupling Facility can be configured to take advan-
capability to have read/write access data from tage of a combination of different Parallel Sysplex
anywhere in the Parallel Sysplex cluster, provide the capabilities:
scalability and availability that businesses demand • Dynamic CF Dispatch: Prior to the availability of the
today. When configured properly, a Parallel Sysplex Dynamic CF Dispatch algorithm, shared CF partitions
cluster has no single point of failure and can provide could only use the “active wait” algorithm. With active
customers with near continuous application availability wait, a CF partition uses all of its allotted time-slice,
over planned and unplanned outages. For detailed whether it has any requests to service or not. The
information on IBM’s Parallel Sysplex technology visit optional Dynamic CF Dispatch algorithm puts a CF par-
our Parallel Sysplex Home Page at ibm.com/servers/ tition to “sleep” when there are no requests to service
eservers/zseries/pso. and the longer there are no requests, the longer the par-
tition sleeps. Although less responsive than the active
The IBM 9037 Sysplex Timer provides a common time
wait algorithm, Dynamic CF Dispatch will conserve CP
reference to all images which assists in managing the
or ICF resources when a CF partition has no work to
process and will make the resources available to other

18 19
partitions sharing the resource. Dynamic CF Dispatch the exploiters, resulting in more reliable subsystem code.
can be used for test CFs and also for creating a hot- •Cost Benefits
standby partition to backup an active CF. Enables the use of non-standalone CFs (e.g. ICFs) for
•Dynamic ICF Expansion. Dynamic ICF expansion pro- all resource sharing and data sharing environments.
vides value by providing extra CF capacity when there •Flexibility
are unexpected peaks in the workload or in case of The diagram below represents creation of a duplexed
loss of CF capacity in the cluster. copy of the structure within a System-Managed CF
– ICF Expansion into shared CPs. A CF partition running Duplexing Configuration
with dedicated ICFs needing processing capacity
beyond what is available with the dedicated CP ICFs, sysplex
timers
can “grow” into the shared pool of application CPs
being used to execute S/390 applications on the same
server.
– ICF Expansion into shared ICFs. A CF partition run-
z/OS ICF ICF z/OS
ning with dedicated ICFs can “grow” into the shared
pool of ICFs in case the dedicated ICF capacity is not
sufficient. The resulting partition, and “L-shaped” LPAR, z900 z900
will be composed of both shared ICF and dedicated
ICF processors, enabling more efficient utilization of
ICF resources across the various CF LPARs. Note: System-Managed CF Duplexing requires CFCC Level
System-Managed CF Structure Duplexing 10 and z/OS 1.2.
System-Managed Coupling Facility (CF) Structure
To understand which of the options and capabilities
Duplexing provides a general-purpose, hardware-
discussed above are suitable for your environment,
assisted, easy-to-exploit mechanism for duplexing CF
please review GF22-5042, Coupling Facility
structure data. This provides a robust recovery
Configuration Options: A Positioning Paper at
mechanism for failure such as loss of a single structure
ibm.com/servers/eservers/zseries/
or CF or loss of connectivity to a single CF, through
library/whitepapers/gf225042.html.
rapid failover to the other structure instance of the
duplex pair.
Parallel Sysplex Coupling Connectivity
Benefits of System-Managed CF Structure Duplexing The Coupling Facilities communicate with z/OS and
include: OS/390 images in the Parallel Sysplex environment
• Availability over specialized high-speed links. For availability
Faster recovery of structures by having the data purposes, it is recommended that there be at least two
already there in the second CF. Furthermore, if a poten- links connecting each z/OS or OS/390 image to each
tial IBM, vendor or customer CF exploitation implemen- CF in a Parallel Sysplex cluster. As processor
performance increases, it is important to also use faster
tation were being prevented by the effort of providing
links so that link performance does not become
alternative recovery mechanism such as structure
constrained. The performance, availability and distance
rebuild, log recovery, etc., that constraint might be
requirements of a Parallel Sysplex environment are the
removed by the much simpler exploitation require-
key factors that will identify the appropriate connectivity
ments for system-managed duplexing.
option for a given configuration.
•Manageability and Usability
A consistant procedure to set up and manage struc- Parallel Sysplex coupling links on the zSeries have
tures across multiple exploiters. been enhanced with the introduction of Peer Mode.
When connecting a z900 server to a z900 Model 100
•Reliability
CF, the links can be configured to operate in Peer Mode.
A common framework provides less effort on behalf of

20 21
This allows for higher data transfer rates to and from the ture is provided on the z900 family of processors. It con-
Coupling Facilities. In Peer Mode, the fiber-optic single sists of a mother card with two daughter cards which
mode coupling link (ISC-3) provides 200 Gb/s capacity, have 2 ports each. Each daughter card is capable of
the ICB-3 link with 1 GB/s peak capacity, and the IC-3 operation at 1 Gb/s in Compatibility Mode or 2 Gb/s in
link with 1.25 GB/s capacity. Additional Peer Mode native mode up to a distance of 10 km. From 10 to 20
benefits are obtained by enabling the link to be MIFed km, an RPQ card is available which runs at 1 Gb/s in
between z/OS (or OS/390) and CF LPARs. The link acts both Peer and Compatibility Modes. The mode is
simultaneously as both a CF Sender and CF Receiver selected for each port via CHPID type in the IOCDS.
link, reducing the number of links required. Larger data The ports are activated in one port increments.
buffers and improved protocols also improve long
• HiPerLinks. HiPerLinks, based on single-mode CF links,
distance performance. For connectivity to 9672s, z900
CF links can be configured to run in Compatibility are available on 9672s and 9674s only. ISC-3s replace
mode with the same characteristics as links on the HiPerLinks on zSeries 900 models.
9672 (single-mode CF links with 100 Gb/s or ICB-3 links • ICB. The Integrated Cluster Bus is used to provide high
with 333 MB/s). All of these above coupling link speeds speed coupling communication between 9672 G5/G6
are theoretical maximums. and/or z900 servers over short distances (~7 meters).
For longer distances, ISC links must be used. Up to 8
Theoretical Maximum Coupling Link Speed ICB links (16 possible via RPQ) are available on the
general purpose z900 models and up to 16 ICB links
Connectivity are available on the z900 Coupling Facility Model 100.
z900 ISC-3 z900 ICB z900 ICB-3
Options ICB-3. The Integrated Cluster Bus-3 is used to provide
high speed coupling communication between two
z900 systems over short distances (~7 meters). For
G2-G6 ISC 1 Gb/s n/a n/a longer distances, ISC-3 links must be used. Up to 16
ICB-3 links are available on both the general purpose
z900 models and the z900 Coupling Facility Model 100.
The performance of the ICB-3 link has been improved
2 Gb/s by higher data rates and new buffering capabilities.
z900 ISC-3 n/a n/a
Peer Mode
IC-3. The Internal Coupling-3 channel emulates the
Coupling links between images within a single server.
G5-G6 ICB n/a 333 MB/s n/a No hardware is required, however a minimum of 2
CHPID numbers must be defined in the IOCDS. IC-3
links provide the fastest Parallel Sysplex connectivity.
1 GB/s Parallel Sysplex Professional Services
z900 ICB-3 n/a n/a
Peer Mode IBM provides extensive services to assist customers
with migrating their environments and applications to
benefit from Parallel Sysplex clustering. A basic set of
IBM services is designed to help address planning and
• ISC-3. InterSystem Coupling Facility-3 channels pro-
early implementation requirements. These services can
vide the connectivity required for data sharing be-
reduce the time and costs of planning a Parallel
tween the Coupling Facility and the systems directly
Sysplex environment and moving it into production. An
attached to it. ISC-3 channels are point-to-point con-
advanced optional package of services is also
nections that require a unique channel definition at
available and includes data sharing application
each end of the channel. ISC-3 channels operating in
enablement, project management and business
Peer Mode provide connection between z900 general
consultation through advanced capacity planning and
purpose models and z900 coupling facilities. ISC-3
application stress testing. For more information on
channels operating in Compatibility Mode provide con-
Parallel Sysplex Professional Services, visit IBM’s Web site
nection between z900 systems and HiPerLink (ISC)
at ibm.com/servers/eserver/zseries/pso/services.html
channels on 9672 Models. A four port ISC-3 card struc-

22 23
Geographically Dispersed Parallel Sysplex processed from a primary VTS, and a copy is stored on
The Geographically Dispersed Parallel Sysplex™ the secondary VTS. Control capability has been added
(GDPS™) complements a multisite Parallel Sysplex to allow GDPS to “freeze” copy operations, so that tape
environment by providing a single, automated solution data consistency can be maintained across GDPS
to dynamically manage storage subsystem mirroring, managed sites during a switch between the primary
processors, and network resources to allow a business and secondary VTSs. Synchronization of system data
to attain “continuous availability” and near transparent sets such as catalogs, the tape control database, and
business continuity/disaster recovery without data loss. tape management databases is also provided with the
GDPS provides the ability to perform a controlled site PtP VTS after an emergency switchover.
switch for both planned and unplanned site outages,
while maintaining full data integrity across multiple Operational data, data that is used directly by
storage subsystems. applications supporting end users, is normally found on
disk. For the past several years, GDPS has provided
GDPS requires Tivoli NetView® for OS/390, System continuous availability and near transparent business
Automation for OS/390, and remote copy technologies. continuity for disk resident data. However, there is
GDPS supports both the synchronous Peer-to-Peer another category of data that “supports” the operational
Remote Copy (PPRC) as well as the asynchronous data, which is typically found on tape subsystems.
Extended Remote Copy (XRC) forms of remote copy. Support data typically covers migrated data, point in
GDPS/PPRC is a continuous availability solution and time backups, archive data, etc. For sustained operation
near transparent business continuity/disaster recovery in the failover site, the support data is indispensable.
solution that allows a customer to meet a Recovery Furthermore, several enterprises have mission critical
Time Objective (RTO) of less than an hour, a Recovery data that only resides on tape. By extending GDPS
Point Objective (RPO) of no data loss, and protects support to data resident on tape, the GDPS solution
against metropolitan area disasters (up to 40 km between provides continuous availability and near transparent
sites). GDPS/XRC is a business continuity/disaster business continuity benefit for both disk and tape
recovery solution that allows a customer to meet a RTO of resident data. Enterprises will no longer be forced to
one to two hours, a RPO of less than a minute, and develop and utilize processes that create duplex tapes
protects against metropolitan as well as regional and maintain the tape copies in alternate sites. For
disasters, since the distance between sites is unlimited. example, previous techniques created two copies of
XRC can use either common communication links and each DBMS image copy and archived log as part of the
channel extender technology between sites or dark fiber. batch process and manual transportation of each set of
tapes to different locations.
Geographically Dispersed Parallel Sysplex
support for Peer-to-Peer Virtual Tape Server Automatic Enablement of CBU for
(PtP VTS): Geographically Dispersed Parallel Sysplex
The GDPS solution has been extended to include tape The intent of the GDPS (CBU) is to enable automatic
data in its management of data consistency and management of the reserved PUs provided by the CBU
integrity across sites with the announced support of the feature in the event of a processor failure and/or a site
Peer-to-Peer VTS configuration (IBM United States failure. Upon detection of a site failure or planned
Hardware Announcement 101-215). The PtP VTS disaster test, GDPS will dynamically add PUs to the
provides a hardware based duplex tape solution and configuration in the takeover site to restore processing
GDPS automatically manages the duplexed tapes in the power for mission-critical production workloads.
event of a planned site switch or a site failure. At the
present time, the GDPS PtP support is only available for a GDPS is discussed in detail in two white papers which
GDPS/PPRC (Peer to Peer Remote Copy) configuration. are available at ibm.com/server/eserver/zseries/pso/
library.html. GDPS is a service offering of IBM Global
A new I/O VTS selection option is provided especially Services. For IBM Installation Services for GDPS refer to
for use with GDPS, so that all virtual volumes are the IBM Web site.

24 25
Continuous Availability IBM ^ zSeries 900
Recommended Configuration

Dedicated (External) IBM's zSeries is the enterprise class e-business server


Coupling Facility
z900 optimized for integration, transactions and data of the
Model 100
9674 or next generation e-business world. In implementing the
9672 R06
new z/Architecture with new technology solutions, the
Internal z900 models are designed to facilitate the IT business
Coupling
Facility transformation and reduce the stress of business-to-
business and business-to-customer growth pressure.
The z900 represents a new generation of servers that
IC
z/OS ICF z/OS feature enhanced performance, support for S/390
Parallel Sysplex clustering, improved hardware
management controls and innovative functions to
address e-business processing.
The z900 has 26 models available as new build
systems and as upgrades from the S/390 Parallel
Enterprise Server™— Generation 5/6. The S/390 R06
ESCON/FICON/FICON Express
Coupling Facility is upgradable to the z900 as well.
The z900 processor enhances performance by
exploiting the new architecture and technology and
Components and assumptions through many design enhancements. The z900 has
•Two Coupling Facilities, at least one external sixteen central processors in a symmetrical processing
•Two Sysplex Timers complex, up from twelve in the S/390 G6. This provides
•Two z/OS or OS/390 servers with redundant backup customers with up to 60 percent performance improvement
capacity over the G6 model ZZ7. At the uniprocessor level
•Two links from each CF to each image performance improvements are in the 20-30 percent range
•Two hardware management consoles when compared to the G6 Model Z17, depending on
•Two ESCON Directors with cross-connected disks workload and environment. The Large System Performance
•Dual electrical power grids Reference (LSPR) should be referenced when considering
•Cloned OS/390 images, latest available software levels performance on the z900. The Web site for LSPR is
•Automation capabilities for recovery/restart ibm.com/servers/eserver/zseries/lspr/.
•Critical data on RAID and/or mirrored disks The I/O subsystem includes Dynamic Channel Path
Management (DCM) and Channel CHPID Assignment.
Key attributes can include These two functions effectively increase the number of
•No single points of failure CHPIDS that can be used for I/O connectivity. DCM
•Fast, automatic recovery: allows channel paths to be dynamically and
– CF: rebuild in surviving CF automatically moved from less utilized devices to
– CEC, z/OS, OS/390: restart subsystems on surviving constrained devices under the supervision of the
image Workload Manager. Channel CHPID Assignment
– TM/DBMS: restart in place permits the assignment of a CHPID to any physical port.
•Surviving components absorb new work This allows the assignment of all 256 CHPIDs to usable
•No service loss for planned or unplanned outages channel paths. Combined, the use of these two functions
•Near unlimited, plug and play, growth capacity allows the full exploitation of the I/O bandwidth inherent
in the 256 channels available on the z900.
For additional information, see the Five Nines/Five Design and technology advances also include introducing
Minutes: Achieving Near Continuous Availability white an improved FIber CONnectivity (FICON) channel card
paper at ibm.com/s390/pso/library. called FICON Express and FICON CTC.
These are some of the significant enhancements in the
zSeries 900 processor that bring improved

26 27
performance, availability and function to the platform. coming from IBM's semiconductor laboratories.
The following sections highlight the functions and Implemented on this chip is the new z/Architecture with
features of the hardware platform. its 64-bit capabilities including new instructions, 64-bit
General Purpose Registers and translation facilities.
z900 Design and Technology
The z900 is designed to provide balanced system The z900 system has a processor storage capability of up
performance. From processor storage to the system's to 64 GB, depending on model. The processor storage is
I/O and network channels, end-to-end bandwidth is implemented in 4, 8 and 16 GB cards. Some z900 models
provided to deliver data where and when it is needed. have 2 memory cards, other z900 models have 4 memory
cards. The recovery design of the memory cards has been
The base of packaging technology for the z900
enhanced with additional spare memory chips.
processor is the MultiChip Module (MCM). The MCM
contains the processor unit (PU) chips, the cache The MultiChip Module is the technology cornerstone
structure chips and the processor storage and I/O for flexible PU deployment in the z900 models. For most
interface chips. The modules measure approximately models the ability of the MCM to have inactive PUs
127 by 127 millimeters. There is one MCM per system allows such features as Capacity Upgrade on Demand
and there are two MCM types. One z900 MCM type (CUoD), the ability to add CPs, ICFs and IFLs
provides support for 20 PUs and 32 MB of level 2 dynamically providing nondisruptive upgrade of
cache making it the world's densest chip module. This processing capability. Also, the ability to add CPs lets a
module contains 101 layers of glass ceramic and 6 small z900 system be a backup for another system in
layers of thin film to provide interconnection between the enterprise; expanding the z900 system to meet an
the chips and the off-module environment. In total, emergency outage situation. This is called Capacity
there is approximately 1 kilometer of wiring on this BackUp (CBU). From a capacity point of view, the MCM
module. This MCM type is positioned for the highest can package up to a full 16-way processor.
performance models. The second MCM type supports 12
PUs and 16 MB of level 2 cache. This module contains 45 The I/O subsystem provides significant bandwidth to
ceramic and 6 thin film layers and is price/performance support the number of processors and the continuing
situated for the lower performance models. advances in high speed I/O. The z900 I/O subsystem
supports up to 96 FICON and/or FICON Express
The design and MCM technology on the z900 provides channels. This represents up to 768 equivalent ESCON
the flexibility to configure the PUs for different uses. channels. (Note that this configuration could also
PUs can be used as Central Processors (CPs), System support 160 ESCON channels {256 CHPIDS total} which
Assist Processors (SAPs), Internal Coupling Facility would represent a combined equivalence of up to 928
processors (ICFs) and Integrated Facility for Linux ESCON channels.) Also, a single zSeries I/O Cage can
processors (IFLs). For the general purpose processor package the maximum of 256 ESCON channels, a feat
models, the number of CPs and SAPs (provide I/O that would have required three I/O cages previously. I/O
processing) are specified by the model number. improvements in the Parallel Sysplex Coupling Link
Optionally, as features on a system, the remaining architecture and technology support faster and more
inactive PUs on the MCM are available as ICF efficient transmission between the Coupling Facility
processors for Coupling Facility applications, IFLs for and production systems. HiperSockets provides a new
Linux applications or as optional SAPs. These high speed capability to communicate between
capabilities are shown, by model, in the Processor Logical Partitions; this is based on high speed TCP/IP
Options table on page 54 and provide the customer memory speed transfers and will provide value in
with tremendous flexibility in establishing the best allowing applications running in one partition to
system for running applications. One PU is always communicate with applications running in another. The
reserved as a spare. z900 Gigabit Ethernet feature is one of the first adapters
in the industry capable of achieving line speed: one
The Processor Unit (PU) uses CMOS 8S with copper
Gigabit per second. Industry standard and openness
interconnect and 47 million transistors with a cycle
are design objectives for I/O in z900. The improved I/O
time of 1.3 nanoseconds for all models. This 17.9 x 9.9
subsystem is delivering new horizons in I/O capability.
millimeter chip is from the latest family of CMOS chips
28 29
Relative Performance of z900 to G5/G6
Servers

The performance design of the z/Architecture enables same server. The number of Linux images that can
the entire server to support a new standard of be supported for productive work is significantly
performance for all applications through expanding more than was previously available due to the bal-
upon a balanced system approach. As CMOS anced system approach. This allows for industry
technology has been enhanced to support not only leading levels of server consolidation for Linux
additional processing power, but also more engines, within the z/Architecture while providing the perfor-
the entire server is modified to support the increase in mance required to support all work types.
processing power. The I/O subsystem supports a great • Support of up to 50% more SAP throughput
amount of bandwidth through internal changes and the - The balanced system approach enables additional
use of FICON, FICON Express and associated CTC, capacity to support a larger number of SAP users
tape, printer and disk devices, thus providing for larger through additional server processing power as well
and quicker data movement into the server. Support of as through the ability to accommodate more users
larger amounts of data within the server required within a single operating system image through the
improved management of storage configurations made new storage hierarchy.
available through integration of the software operating • Support of up to 50% more Lotus® users
system and hardware support of 64-bit. New coupling - The balanced system approach enables additional
links in the architecture allow customers to continue to capacity to support a larger number of Lotus users
exploit the advantage of the leading clustering solution in through additional server processing power as well
the industry, Parallel Sysplex technology. The combined as the ability to accommodate more users within a
balanced system effect allows for increases in single operating system image through the new stor-
performance across a broad spectrum of work. age hierarchy.
• Support of up to 45% more throughput in Web-initiated
When comparing a 16-way zSeries Model 116 to a G6 e-business transactions
turbo ZZ7, the following levels of performance can be - The unique performance characteristic of the
obtained: z/Architecture is that the balanced system approach
• Savings in batch elapsed time of up to 17% for the not only enables additional transactions to be ex-
CPU component of an individual batch job. A batch ecuted but it is also a design point that is well suited
stream can see up to a 60% throughput improvement to contain and manage spikes in work based upon
with a potential for further savings via implementation an influx of Web-initiated transactions. This capability
of FICON, compression, and ESS DASD. is achieved through a combination of the operating
- Due to the increase in hardware capacity, additional system as well as increased throughput delivered
processing power can be used to reduce the CPU with the z900 server.
component of the batch job. Combined with the sup-
Note: Expected performance improvements are
port of larger data movement, this allows for improved based on hardware changes. Additional performance
turnaround time in CPU constrained batch jobs. benefits may be obtained as the z/Architecture is
• Ability to support 55% additional Linux images fully exploited.
- The z/Architecture not only supports traditional work
but also allows Linux work to execute within the

9672 G5 R Models 9672 G5 Y Models 9672 G6 X Models 9672 G6 Z Models z900 Capacity z900 Models 101-116
Models 1C1-1C9
30 31
z900 Family Models

The z900 has a total of 26 models to offer flexibility in with the Models 101-109, has a cycle time of 1.3
selecting a system to meet the customer’s needs. nanoseconds. The Models 110-116 can easily upgrade
Twenty-five of the models are general purpose systems. from one model to the next through Capacity Upgrade
The remaining model is the Coupling Facility Model on Demand and support Capacity Backup (CBU). The
100. There are a wide range of upgrade options Models 110-116 have 3 SAPs as standard and up to 24
available which are described below and shown on STI links for I/O attachment. Any of the spare PUs on the
the following pages. Capacity Upgrade on Demand MCM can be assigned as a CP, SAP, ICF, or IFL.
and Capacity Backup (CBU) are available. The z900
has also been designed to offer a high performance z900 Models 1C1-1C9
and efficient I/O structure to meet the demands of These nine models are general purpose systems and
e-business as well as high demand transaction range from a 1-way to a 9-way symmetrical
processing applications. Up to 256 ESCON channels multiprocessor. The systems have a design and cycle
will now fit into a single I/O cage; or a total of 96 FICON
and/or FICON Express channels and 160 ESCON
channels can be accommodated in a fully configured z900 Models
system. To provide the best choice of processor for the
application, two designs have been developed.
Although similar in structure, one design has a 12-
processor unit (PU) MulitChip Module (MCM) and two
memory cards. The other design has a 20 PU MCM and
4 memory cards. Both have equivalent I/O capability.
The processor models and a discussion of
configurations follow.

z900 Models 101-109


These nine models are general purpose systems and
range from a 1-way to 9-way symmetrical
multiprocessor (SMP). The systems have a 12-PU MCM,
two memory buses, and can support up to 32 GB
processor storage (entry storage is 5 GB). The PU has a
cycle time of 1.3 nanoseconds (ns). These models can
easily upgrade from one model to the next through
Capacity Upgrade on Demand. Also, there are upgrade
paths to the Models 110 through 116; however, this
upgrade will require a system outage. The Models 101-
109 have 2 System Assist Processors (SAPs) as
standard and up to 24 STI links for I/O attachment. Any of
the spare PUs on the MCM can be assigned as a CP, SAP,
Integrated Coupling Facility (ICF) or Integrated Facility for
Linux (IFL).

z900 Models 110-116


These seven models are general purpose systems and
range from a 10-way to a 16-way symmetrical
multiprocessor (SMP). The systems have the 20-PU
MCM, four memory buses and can support up to 64 GB
processor storage (entry storage is 10 GB). The PU, as

32 33
time identical to the models 110-116 including the 20 PU same number of channels. For FICON and FICON
MCM, four memory buses and supporting up to 64 GB Express, the nI/O cage can accommodate up to 16
processor storage (entry storage is 10 GB). The Models cards or 32 channels per cage; with the previous
1C1-1C9 are available as an option for Capacity technology, up to 36 channels would require three I/O
Upgrade on Demand and Capacity Backup (CBU) cages. Thus, much denser packaging and higher
requirements and can be upgraded to a 16-way z900 bandwidth has been achieved.
without a system outage. Customers whose capacity
requirements are likely to exceed the Model 101-109 The Z-Frame is an optional I/O Frame and attaches to
range should consider the 1C1-1C9 as an alternative. the A-Frame. The Z-Frame can contain up to two of the
new zSeries I/O (nI/O) cages, up to two compatibility I/O
z900 Coupling Facility Model 100 (cI/O) cages or a mixture of both. Following is the layout
The Model 100 is the standalone Coupling Facility in of the A- and Z-Frames and both types of I/O cages.
the z900 family. This Model can have up to 9 ICF
I/O Car
Car d s: The next topic has a chart that shows cards
ard
engines. It is recommended that the z900 CF Model
supported by the two types of I/O cages.
100 be used in production data sharing configurations
for its improved coupling efficiency.
Up to 256 ESCON Channels
Customers can upgrade current 9672 R06 Models to Two different ESCON channel cards are available on
the z900 Coupling Facility Model 100, maximizing the the z900 system:
coupling efficiency. The z900 CF Model 100 can be • A new 16 port channel card which plugs into the new
upgraded to the z900 general purpose models. I/O cage will be used for all new ESCON channel or-
ders and any overflow from Compatibility I/O cages on
I/O Connectivity G5/G6 upgrades. Up to 15 ports will be used for ESCON
The z900 contains an I/O subsystem infrastructure connectivity; one port is reserved as a spare port.
which uses an I/O cage that provides 28 I/O slots • Four port channel cards which plug into the Compatibility
compared to the G5/G6 style cage with 22 slots. I/O cage are used only in upgrades from G5/G6 to z900.
ESCON, FICON, FICON Express and OSA-Express cards • Channels are available in four port increments for both
plug into the zSeries I/O (nI/O) cage. Parallel channel, card types.
OSA-2 Token-Ring and FDDI and ESCON 4-port cards
are accommodated in a Compatibility I/O (cI/O) cage
(22 slots). ESCON 4-port cards are used only in Cage Layout and Options
upgrading from a G5/G6 model. All I/O cards and their
support cards can be hot-plugged in the nI/O cage.
Installation of an I/O cage remains a disruptive MES so A-Frame Z-Frame A-Frame
the Plan Ahead feature remains an important B B
consideration when ordering a z900 system. - -
CEC nI/O or
F CEC F
The nI/O cage takes advantage of an exclusive IBM r cI/O Cage r
a a
packaging technology that provides a subsystem with
nI/O Cage
m nI/O or nI/O Cage m
approximately seven times higher bandwidth than the e e
previous G5/G6 I/O cage. Each general purpose z900 cI/O Cage only
model comes with one nI/O cage standard in the
A-Frame (the A-Frame also contains the processor
CEC cage). The nI/O cage, using new 16 port ESCON
cards, can hold 256 ESCON channels; previous
packaging required three I/O cages to package the

34 35
Up to 96 FICON or FICON Express Channels Eighty-eight (88) parallel channels can be ordered on a
The z900 supports up to 96 FICON and/or FICON new-build z900 and up to 96 via RPQ 8P2198. This RPQ
Express channels. Both FICON and FICON Express are provides an additional Compatibility I/O cage to enable
available in long wave (LX) and short wave (SX) installation of the extra parallel channel cards. This RPQ
operation. Each FICON and FICON Express card has is not required if a G5/G6 Model with greater than 88
two channels per card. The LX and SX cannot be parallel channels is upgraded to a z900.
intermixed on a single card. The maximum number of
FICON cards is 48 installed in 3 nI/O cages. ISC-3
A four port ISC-3 card is provided on the z900 family of
Up to 88 Parallel Channels processors. It consists of a mother card with two
The four port parallel channel card is the same card daughter cards which have two ports each. Each
used on G5/G6 models and is orderable on the z900. daughter card port is capable of operation at 1 Gb/s in
However, it must plug into a Compatibility I/O cage. The Compatibility Mode or 2 Gb/s in native mode up to a
three port parallel card, if present during an upgrade distance of 10 km. The mode is selected for each port
from a G5/G6, will be carried forward. via CHPID type in the IOCDS. The ports are orderable in
one port increments.
zSeries Compatibility An RPQ card (8P2197) is available to allow ISC-3
I/O Type
I/O Cage I/O Cage distances up to 20 km. This card runs in Peer Mode at
1 Gb/s and/or Compatibility Mode at 1 Gb/s. The ports
ESCON Yes (16 port) Yes (4 port/MES) are orderable in two port increments.

Parallel Channel CHPID Assignment


No Yes
Channel The z900 provides customers with the option of
mapping CHPID numbers to physical channel ports.
ISC-3 (1 & 2 This is called Channel CHPID Assignment. Channel
Yes No
Gigabit) CHPID Assignment helps customers maintain G5/G6
I/O definitions during system upgrades and in
FICON
Yes No establishing new I/O definitions. It also allows for a
FICON Express
logical sequence of CHPID numbers to be assigned to
Fast Ethernet Yes No a control unit. A CHPID Mapping tool is available on
Resource Link, ibm.com/servers/resourcelink, to assist
ATM 155 Yes No with mapping.

PCICC Yes No

PCICA Yes No

Token-Ring Yes No

GbEthernet Yes No

FDDI No Yes

OSA-2
No Yes
Token-Ring

36 37
Cryptography

IBM leads the industry in bringing greater security to The CUoD functions are:
e-business with its high availability CMOS • z900 continues to support the dynamic CUoD function
Cryptographic Coprocessors. This feature has earned introduced on G5/G6.
Federal Information Processing Standard (FIPS) 140-1 - Nondisruptive CP, ICF, and IFL upgrades are available
level 4, the highest certification for commercial security within minutes.
ever awarded by the U.S. Government. For the z900, the • Dynamic upgrade of all I/O cards in the nI/O Cage.
two Cryptographic Coprocessors Single Chip Modules • Dynamic upgrade of spare installed memory.
(SCMs) have been moved from the MCM to the CPC
Cage. The SCMs are plugged directly into the rear of Plan Ahead and Concurrent Conditioning
the CPC backplane. The SCMs are individually Concurrent Conditioning configures a system for hot
serviceable, minimizing system outage. plugging of I/O based on a future specified target
configuration. Concurrent Conditioning of the zSeries
The z900 servers can also support a combination of up
I/O is minimized by the fact that all I/O cards, plugging
to 8 optional Peripheral Component Interconnect
into the zSeries I/O (nI/O) cage, are hot pluggable. This
Cryptographic Coprocessor (PCICC) or the PCI
means that the only I/O to be conditioned is the nI/O
Cryptographic Accelerator (PCICA) features. Each
cage itself. The question of whether or not to
PCICC/PCICA feature contains 2 cryptographic
concurrently condition a cage is a very important
coprocessor accelerators, for a maximum of 16
consideration especially with the rapid change in the IT
coprocessor accelerators. This provides the capability
environment (e-business) as well as the technology.
to support up to 3800 SSL transactions/second.
Migration to FICON, FICON Express or additional OSA-
The combination of the two coprocessor types enables Express networking is made exceptionally easy and
applications to invoke industry-standard cryptographic nondisruptive with the appropriate microcode load and
capabilities — such as DES, Triple DES, or RSA — for if the cage space is available. Planning ahead is an
scalable e-transaction security and the flexibility to important activity when purchasing a typical new build
adopt new standards quickly. system (which will probably have one nI/O cage, one
cI/O cage and space for one additional nI/O cage in the
The new PCICA feature is designed to address the Z-Frame). This is even more important when
high performance SSL needs of e-business considering a G5/G6 system upgrade where the
applications, and has a design point different from the Z-Frame space is occupied with cI/O cages. FICON,
existing zSeries CMOS Cryptographic Coprocessor FICON Express, OSA-Express, Crypto PCICC and ISC-3
and zSeries PCICC feature. cards plug only into nI/O cages.

z900’s Capacity Upgrade on Demand (CUoD) The z900 now supports concurrent memory upgrade.
Capacity Upgrade on Demand allows for the This capability will allow a processor's memory to be
nondisruptive addition of one or more Central increased without disrupting the processor operation. To
Processors (CPs), Internal Coupling Facilities (ICFs) take advantage of this capability, a customer should not
and/or Integrated Facility for Linux (IFLs). Capacity plan processor storage on the 8, 16 or 32 GB
Upgrade on Demand can very quickly add processors increments. A customer with a Model 106, for example,
up to the maximum number of available inactive with 5 GB of storage will be able to concurrently
engines. This provides customers with value for much upgrade to 6, 7 and 8 GB but will not be able to get to
needed dynamic growth in an unpredictable the next increment of 10 GB without a disruption.
e-business world. The Capacity Upgrade on Demand
The Plan Ahead process can easily identify the
function combined with Parallel Sysplex technology
customer configuration that is required to meet future
enables virtually unlimited capacity upgrade capability.

38 39
needs. The result of concurrent conditioning is a needs of the larger CBU target machine. When capacity
flexible IT infrastructure that can accommodate is needed in an emergency, the primary operation
unpredictable growth in a low risk, nondisruptive way. performed is activating the emergency CBU
Depending on the required Concurrent Conditioning, configuration with the reserved PUs added into the
there should be minimal cost associated with dormant configuration as CPs.
z900 capacity. This creates an attractive new option for
businesses to quickly respond to changing Upon request from the customer, IBM can remotely
environments, bringing new applications online or activate the emergency configuration. This is a fast
growing existing applications without disrupting users. electronic activation that eliminates time associated
with waiting for an IBM CE to arrive onsite to perform
z900 Server Capacity BackUp (CBU) the activation. A customer request through the
Capacity BackUp (CBU) is offered with the z900 Hardware Master Console and Remote Support Facility
processors to provide reserved emergency backup could drive activation time down to minutes; a request
CPU capacity for situations where customers have lost by telephone (for customers without RSF) could drive
capacity in another part of their establishment and activation to less than an hour.
want to recover by adding reserved capacity on a The z900 supports concurrent CBU downgrade. This
designated z900 system. A CBU system normally function enables a Capacity Backup Server to be
operates with a "base" CPU configuration and with a returned to its normal configuration without an outage
preconfigured number of additional Processor Units (i.e. Power-on reset).
(PUs) reserved for activation in case of an emergency.
Automatic Enablement of CBU for Geographically
The z900 technology is ideally suited for providing
Dispersed Parallel Sysplex (GDPS)
capacity backup since the reserved CBU processing
The intent of the GDPS CBU is to enable automatic
units are on the same technology building block, the
management of the reserved PUs provided by the CBU
MCM, as the regular CPs. Therefore, a single processor
feature in the event of a processor failure and/or a site
can support two diverse configurations with the same
failure. Upon detection of a site failure or planned
MCM. For CBU purposes, the Models 101-109 can scale
disaster test GDPS will dynamically add PUs to the
from a 1-way to a 9-way; the Models 110-116 from a 10-way
processors in the takeover site to restore processing
through a 16-way; and the Models 110-116 in conjunction
power for mission-critical production workloads.
with the Capacity models (1C1-1C9) can scale from
1-way through 16-way with the purpose of providing
capacity backup.

The "base" CBU configuration must have sufficient


memory and channels to accommodate the potential

40 41
S/390 Fibre (Channel) CONnection (FICON)

FICON available in increments of one and the z900 FICON


The e-business world is demanding. With the and FICON Express channels are available in
requirements for fast data access, continuous data increments of two. On the z900 server, a maximum of
availability, improved flexibility — all with lower cost of 32 channels (16 cards) can be plugged into a single
ownership — zSeries with native FICON-attached new z900 I/O cage, with a total of 96 channels (48
devices can help achieve these goals. cards) in three cages.
Native FICON delivers leading edge high speed data To be consistent with the requirement to provide an
transfer technology to servers, switches, control units, “OPEN” platform for connectivity, FICON and FICON
and storage devices. FICON and FICON Express Express are compatible with the Fibre Channel
channels allow a more efficient and faster data transfer Physical and Signaling Standard (FC-PH) and are in
while, at the same time, allowing customers to use their the approval process to NCITS for formal adoption into
currently installed single mode and multimode fiber the FC-4 layer of the Fibre Channel standard under the
optic cables. FICON and FICON Express channels are name FC-SB-2.
available with both longwave and shortwave options.
FICON and FICON Express support a maximum
FICON and FICON Express unrepeated distance of up to 10 km (6.2 miles) (20 km
FICON was first introduced on the G5 servers which via RPQ, up to 100 km with repeaters) for nine micron
support a maximum of 24 FICON channels. The G6 single mode fiber and up to 550 meters (1,804 feet) for
servers support a maximum of 36 FICON channels, and 50 or 62.5 micron multimode fiber through the use of
the z900 servers support a maximum of 96 FICON long wavelength FICON channel cards and Mode
channels in any mix of long wavelength and short Conditioning Jumper (MCP) cables. FICON and FICON
wavelength features. A single FICON channel is Express use different jumper cables which are
capable of supporting up to 3600 I/O operations per supplied as required during configuration/ordering
second on G5/G6 and up to 5000 I/O operations/sec on process. The maximum unrepeated distances for the
z900. A FICON Express channel can support up to short wavelength FICON and FICON Express channel
7200 I/O operations/sec in native mode on z900. cards are up to 500 meters for 50 micron multimode
fiber and 300 meters (984 feet) for 62.5 micron
With its new internal bus and LC connectors, FICON multimode fiber. FICON and FICON Express are
Express is the z900’s latest implementation of the Fibre designed to reduce the data droop effect that made
Channel Architecture. Available only on the zSeries, this long distances not viable for ESCON. This distance
card has two channels and can achieve improved capability is becoming increasingly important as
performance over the previous generation FICON customers are moving toward remote I/O, vaulting for
channel card. The z900 servers can support a disaster recovery and Geographically Dispersed
maximum of 96 FICON Express channels in any mix of Parallel Sysplexes for availability.
longwave and shortwave features. A single FICON
Express channel is capable of supporting up to 7200 FICON Bridge
I/O operations/sec on z900. Customers can leverage The 9032 Model 005 ESCON Director supports FICON
this additional capacity to consolidate channels and and FICON Express longwave features through the use
reduce configuration complexity, infrastructure costs, of a bridge card and is intended to help provide
and the number of channels that must be managed. investment protection for customers’ currently installed
ESCON control units. Up to 16 bridge cards are
A FICON channel can coexist with ESCON and parallel supportable on a single 9032 Model 005 with each
channels in any model S/390 G5/G6 or z900 server and card capable of sustaining up to eight concurrent
is supported by Multiple Image Facility (MIF). A FICON ESCON data transfers. Current 9032 Model 005 ESCON
Express channel can coexist with ESCON and FICON Directors can be field upgradable at no charge to
channels in any model z900 server and is supported support the FICON bridge cards, and both bridge cards
by MIF as well. The G5/G6 FICON channels are and ESCON cards can coexist in the same box.

42 43
Native FICON FICON Connectivity
Native FICON support is available on G5/G6 and G5/G6 ESCON
zSeries servers. CU
FICON Bridge
Currently, the IBM ESS models F10 and F20 have two
new host adapters to support native FICON. These host ESCD ESCON
9032 CU
adapters each have one port per card and can either
Model 5
be FC 3021 Fibre Channel/FICON (long wave) Host
Adapter or FC 3032 Fibre Channel/FICON (short wave) ESCON
Host Adapter. The F10 and F20 models can support up CU
FICON Bridge
to 16 FICON ports per ESS.
ESS
The IBM Magstar® 3590 A60 Tape controller provides z900 F10, F20
up to two FICON interfaces which can coexist with
ESCON on the same box. Customers can utilize IBM’s T
POR
r 64
highest capacity, highest performance tape drive to 32 o MAGSTAR
support their new business models. 2032 3590 A60

Four Fibre Channel directors are now available to


provide dynamic connectivity to native FICON control
units. The IBM 2032 models 001 and 064 T
POR
(resells of the McDATA ED-5000 and ED-6064) are 32 r 128
64 o
and 64 port high availability directors. The IBM 2042
All FICON Channels = 100MB/s
2042
models 001 and 128 (a resell of the Inrange FC/9000)
MAGSTAR
are 64 and 128 port high availability directors. All have = LX ONLY
3590 A60
features that provide interface support to allow the unit = LX or SX
to be managed by System Automation for OS/390.
ESS
OS/390 Resource Measurement Facility (RMF™) reports F10, F20
on the bus utilization, as well as port bandwidth for
performance management and capacity planning. FICON CTC
OS/390 R3 or higher with PTFs is required for FICON FICON and FICON Express on z900 now support CTC.
and FICON Express and associated RMF support in the G5 and G6 servers can connect to z900 FICON CTC.
9032 Model 005 FICON bridge environment and This channel to channel connectivity will increase
OS/390 R6 or higher with PTFs is required for non- bandwidth between G5, G6 and z900 systems.
bridge environments.
In addition, FICON CTC on the z900 is not limited to
Native FICON channels will help to reduce bandwidth intersystem connectivity (as is the case with ESCON),
constraints and channel contention to enable easier but will also support multiple device definitions. ESCON
server consolidation, new application growth, large channels that are dedicated as CTC cannot
business intelligence queries and exploitation of communicate with any other device. z900 FICON and
e-business. FICON Express channels are not dedicated to CTC
only. They can support both device and CTC mode
definition concurrently, allowing for greater connectivity
flexibility.

44 45
Open Systems Adapter-Express Features
(OSA-Express)

The Open Systems Adapter-Express (OSA-Express) z900 OSA-Express Fast Ethernet


features (Gigabit Ethernet, Fast Ethernet, and 155 ATM,
and the new Token-Ring) have been redesigned to non-QDIO mode QDIO Mode - TCP/IP
SNA Passthru
support the new I/O infrastructure of zSeries. These TCP/IP Passthru IBM

HPDT MPC
features combined with z/OS or OS/390, z/VM or VM/ESA,
RS 6000

IBM

Linux and VSE/ESA in the future, deliver a balanced Netfinity


pSeries, RS/6000

system solution to maximize throughput and minimize host xSeries, Netfinity

Switch/ 10/100 Mbps


interrupts to continue to satisfy your business goals. The Hub/ Switch/ Ethernet
Router Hub/
redesigned zSeries OSA-Express feature requires the new Router

I/O Cage.
10/100 Mbps
The OSA-Express features currently available on the S/390 Ethernet

Parallel Enterprise Servers G5 and G6 are not available on Server Server

z900. A Feature Conversion is available to help G5/G6 IP Router

Gigabit Ethernet, Fast Ethernet, ATM and Token-Ring DLSw Router


Internet or
users to move z900 OSA-Express. Users of OSA-2 extranet
IP WAN
FDDI will require a compatibility I/O cage on the z900. IP WAN
Intranet Intranet

Remote Office Remote Office

A higher performing Gigabit Ethernet feature with a 66 4/16 Mbps


Token-Ring
4/16 Mbps
Token-Ring
MHz, 64-bit PCI bus capable of achieving line speeds of DLSw
Router
IP Router
1 Gb/s is available on the z900. This new design
incorporates two ports of connectivity in a single I/O slot. TCP/IP applications
TN3270 browser
Each port uses one CHPID. These features attach to the SNA DLSw
TCP/IP
access to SNA
Enterprise Extender
STI bus at 333 MB/s. Native SNA for SNA end points

10/100 Mbps 10/100 Mbps


Server Server
Ethernet Ethernet
The infrastructure for Fast Ethernet and 155 ATM and
the new Token-Ring available on the z900 offers two
ports of connectivity in a single I/O slot. The Fast
Ethernet runs at either 10 or 100 Mbps. The ATM runs at
155 Mbps. The new Token-Ring runs at either 4 or 16
and now 100 Mbps permitting connection to higher Queued Direct Input/Output (QDIO)
speed Token-Ring switch backbones and servers. Each The OSA-Express Gigabit Ethernet, Fast Ethernet, and
OSA-Express port uses one CHPID. These features 155 ATM (Ethernet LAN Emulation), and Token-Ring
also attach to the STI bus at 333 MB/s. features support QDIO, which was first introduced in
Communication Server for OS/390 Version 2 Release 7.

Queued Direct Input/Output (QDIO), a highly efficient


data transfer architecture, breaks the barriers associated
with the Channel Control Word (CCW/ESCON)
architecture increasing data rates and reducing CPU
cycle consumption. QDIO is designed to dramatically
reduce system overhead and improves throughput
using S/390 memory queues and a signaling protocol
to directly exchange data between the OSA-Express
microprocessor and CS for OS/390 R10 and z/OS.

46 47
z900 OSA-Express Gigabit Ethernet z900 OSA-Express Token-Ring
QDIO Mode - TCP/IP
non-QDIO mode QDIO Mode - TCP/IP
SNA Passthru
IBM

RS 6000

pSeries, RS/6000 TCP/IP Passthru


HPDT MPC 4/16 Mbps
Token Ring
4/16/100 Mbps
Switch/ Token Ring
IBM Hub/ Backbone
xSeries, Netfinity Router Server
Netfinity
100 Mbps 100 Mbps
Server Token Ring Token Ring
Backbone
Switch/
Hub/
Router
4/16/100 Mbps
Token-Ring 4/16 Mbps 4/16 Mbps
Token Ring Token Ring
Switch/ Switch
Hub/ Router
Router

Server 16/100 Mbps Server


Server Token Ring
IP Router

Gigabit 4/16 Mbps


Token Ring Internet or
Ethernet DLSw Router extranet
Server
IP WAN IP WAN
Intranet Intranet

Server Remote Office Remote Office


4/16 Mbps 4/16 Mbps
10/100 Mbps IP Router Token-Ring Token-Ring
Ethernet DLSw
Router
IP Router

IP WAN Internet or
TN3270 browser
Intranet extranet access to SNA
SNA DLSw Enterprise Extender
TCP/IP for SNA end points
Native SNA TCP/IP applications
TCP/IP applications Remote Office
TN3270 browser 10/100 Mbps 10/100 Mbps
Server Server
Ethernet Ethernet
access to SNA
IP Router
Enterprise Extender
for SNA end points

4/16 Mbps Server to User connections


Token-Ring
Server A key strength of OSA-Express and associated
Communications Server protocol support is the ability
to accommodate the customer’s attachment
requirements, spanning combinations of TCP/IP and
NON-QDIO operational mode SNA applications, and devices. Customers requiring
The OSA-Express Fast Ethernet, the new Token-Ring, TCP/IP connections from the remote site to the TCP/IP or
and the 155 ATM (native, Ethernet and Token-Ring LAN SNA applications on z900 and S/390 can use OSA-
emulation) features also support the non-QDIO mode of Express with QDIO and either direct TCP/IP access or
operation. The adapter can only be set (via the CHPID appropriate SNA to IP integration technologies such as
type parameter) to one mode at a time. The non-QDIO, TN3270 Server and Enterprise Extender. Customers
mode does not provide the benefits of QDIO. However it who require the use of SNA-based connections from
does provide for protocol support similar to the OSA-2 the remote site can use a TCP/IP or SNA transport to the
adapter but at higher levels of performance. This data center and connection into z900 and S/390 using
support includes native SNA, APPN®, High Performance OSA-Express in non-QDIO mode.
Routing, TCP/IP passthru, and HPDT MPC.

48 49
LPAR Support of OSA-Express Open Systems Adapter 2
The OSA-2 Ethernet-Token-Ring adapter feature #5201
For customers who use the Processor Resource/ (sometimes called the ENTR or OSA-2 Token-Ring) will
Systems Manager™ (PR/SM™) capabilities of the z900 not be available on the z900 with the availability of the
and the S/390 servers, IBM offers the G5 and G6 new OSA-Express Token-Ring adapter.
Multiple Image Facility (MIF), allowing the sharing of
physical channels by any number of logical partitions The Open Systems Adapter 2 (OSA-2) Fiber
(LPARs). Since a port on an OSA-Express feature is like Distributed Data Interface (FDDI) feature continues
a channel, sharing of an OSA-Express port is done to be supported on the z900, and is available for
using MIF. The LPARs are defined in the Hardware new builds and for upgrades.
Configuration Definition (HCD), and each LPAR is given
a specific partition ID. Depending upon the feature, and For planning purposes, the OSA-2 Token-Ring and
how it is defined, SNA/APPN/HPR and TCP/IP traffic can FDDI feature can only be used in the compatibility
flow simultaneously through any given port. I/O cage, while the new OSA-Express Token-Ring
uses the new I/O cage.
z900 OSA-Express Asynchronous Transfer Mode The OSA-2 feature require the S/390 Open Systems
Adapter Support Facility (OSA/SF) when configuring
and customizing the feature, and updating the
ATM Native supported software.
155 Mbps
ATM Note: OSA-Express Token-Ring can now be used in-
stead of OSA-2 Token-Ring, removing the requirement
for the cI/O cage slot.
ATM
Network
Server

ATM LANE

155 Mbps
ATM
LE Network

Switch/ Server
Switch/
Hub/ Hub/
Router Router

LAN 10/100 Mbps Ethernet


Emulation
Server

Server

4/16 Mbps
Token-Ring

50 51
Advanced Availability Functions

Transparent Sparing Cryptographic Coprocessors: The Cryptographic


z900 offers two MCMs, a 20-processor version (max 16 Coprocessors are designed as Single Chip Modules
Central Processors) and a 12-processor version (max 9 mounted on the Processor Board and are individually
CPs). The added capacity is availability for nondisruptive serviceable. This eliminates the need to change the
customer growth or, in the case of processor failure, for MCM (where they previously resided) in the event of a
transparent sparing. crypto chip failure, reducing downtime.

Enhanced Dynamic Memory Sparing Concurrent Service for I/O Cards: All the cards which
The z900 has enhanced this robust recovery design plug into the new I/O Cage are able to be added and
with 16 times more chips available for sparing. This will replaced concurrent with system operation. This
virtually eliminate the need to replace a memory card eliminates any need to schedule outage to service or
due to DRAM failure. upgrade the I/O subsystem on this cage.

Enhanced Storage Protect Keys: z900 has enhanced Redundant Cage Controllers: The Power and Service
the memory storage protect key design by adding a Control Network features redundant Cage Controllers
third key array to each memory card. The arrays are for Logic and Power control. This design enables
parity checked and employ a Triple Voting strategy to nondisruptive service to the controllers and eliminates
assure accuracy. This will reduce the need for memory customer scheduled outage.
card replacement due to key array failure.
Auto-Switchover for Service Element: The z900 has two
ESCON Port Sparing: The ESCON 16-port I/O card is Service Elements. In the event of failure on the Primary
delivered with one unused port dedicated for sparing SE, the switchover to the backup is handled
in the event of a port failure on that card. Other unused automatically. There is no need for any intervention by
ports are available for nondisruptive growth of ESCON the customer or Service Representative.
channels.
Concurrent Capacity Backup Downgrade
Concurrent Maintenance This function allows the customer to downgrade the
Upgrade for Coupling Links: z900 has concurrent disaster backup machine to its normal configuration
maintenance for the ISC-3 adapter card. Also, Coupling without requiring the PowerOn Reset (POR).
Links can be added concurrently. This eliminates a
need for scheduled downtime in the demanding Concurrent Memory Upgrade
sysplex environment. This function allows adding memory concurrently up to
the maximum amount physically installed.

52 53
zSeries 900 Exclusive Functions

- z/Architecture zSeries 900 Model Upgrades


64-Bit Architecture • Comprehensive upgrade paths available
- Intelligent Resource Director
• G5 to z900 Models
- LPAR CPU Management
- Dynamic Channel Path Management • G6 to z900 Models
- Channel Subsystem Priority Queuing • G5 R06 Coupling Facility to z900 Model 100 CF
- HiperSockets • G5 R06 CF to z900 Models
- Internal System Channel-3 • Vertical upgrades within z900
- Integrated Cluster Bus-3 • Vertical upgrades within z900 Model 100 CF
- Internal Coupling Channel-3
• z900 Model 100 CF to z900 Models
- Channel CHPID Assignment
- Nondisruptive replacement of I/O
- Two port OSA-Express adapters at line speed
- ESCON Port Sparing
- FICON Express
- FICON CTC
- Concurrent Maintenance for ISC-3 adapter card
- Concurrent upgrade for Coupling Links
- Concurrent Service for all I/O cards
- Total 512K subchannel addresses
- Redundant Cage Controllers
- Auto-Switchover for Service Element
- More Dense Logic Modules (MCMs)
- Enhanced Storage Protect Keys
- Enhanced Dynamic Memory Sparing
- Enhanced Hardware Compression
- Cryptographic Coprocessors (Single Chip Module)
- Integrated Facilities for Linux (IFLs)
- Nondisruptive CBU downgrade
- Concurrent memory upgrade
-Note:
-- G5/G6 supports IFL
-- z900 supports all G5/G6 functions, except
> Asynchronous Data Mover Facility (ADMF)
> Integrated Coupling Migration Facility (ICMF)

z900 Capacity z900 Models 101-116


Models 1C1-1C9

54 55
Processor Options z900 Models 101-116, 1C1-1C9 zSeries 900 Family Configuration Detail

Total Comments
I/O Channel / Compatibility
PUs CPs SAPs IFLs*/ ISC-3* STIs ports or I/O Cage I/O Cage
Quantity Channel Total
ICB-3* ICB*** min 1 - max 3 2 of Increments I/O cages
Std Opt* ICFs* STIs/Feature1 min 0 - max 23
Channels4 min 1 - max 35

ESCON
101 12 1 2 3 8 16 8** 32 24 - 4 port
No Yes 176 / 44 4 G5/G6 upgrades only

102 12 2 2 3 7 16 8** 32 24 ESCON


Yes No 256 / 18
4 (LIC
1 spare port per card
- 15/16 port CC)
103 12 3 2 3 6 16 8** 32 24 Parallel 96 via RPQ or G5/G6
No Yes 88 / 22 4
- 4 port u/g
104 12 4 2 3 5 16 8** 32 24
Parallel
No Yes 96 / 32 3 MES upgrade only
105 12 5 2 3 4 16 8** 32 24 - 3 port
FICON/FICON
106 12 6 2 3 3 16 8** 32 24 Express Yes No 96 / 48 2 —
- 2 port7
107 12 7 2 2 2 16 8** 32 24
PCICC
Yes No 16 / 8 2 Not defined in IOCP
108 12 8 2 1 1 16 8** 32 24 - 2 processors

PCICA
109 12 9 2 0 0 16 8** 32 24 Yes No 12 / 6 2 —
- 2 processors

110 20 10 3 5 6 16 8** 32 24 Fast Ethernet


Yes No 24 / 12 2 G5/G6 upgrades only
- 2 port
111 20 11 3 5 5 16 8** 32 24 Gbit Ethernet7
Yes No 24 / 12 2 Different features
- 2 port
112 20 12 3 4 4 16 8** 32 24 7
155 ATM
Yes No 24 / 12 2 Different features
113 20 13 3 3 3 16 8** 32 24 - 2 port

FE, Gbit and 24 ports (12


114 20 14 3 2 2 16 8** 32 24 — — — —
ATM card)

115 20 15 3 1 1 16 8** 32 24 Token Ring Yes No 24 / 12 2 G5/G6 upgrades only

116 20 16 3 0 0 16 8** 32 24 FICON, FICON


48 card/system
Express, PCI-
— — — — 16 card/
CC/CA, FE,
I/O cage
GbE, ATM, TR
9 OSA-2
100 12 0 2 0 16 16 32**** 24 FDDI + TR = 12
(ICF) - TR - FDDI - 1 No Yes 12 / 12 1
cards max
port
OSA-2 FDDI + TR = 12
No Yes 24 / 12 1
- TR - 2 ports cards max
1C1 20 1 3 5 15 16 8** 32 24
ISC-3 20 km
1 (LIC
1C2 20 2 3 5 14 16 8** 32 24 (1 & 2 Gbit) Yes No 32 / 8 via RPQ
CC)
-2 + 2 ports (H/W 1Gb)
1C3 20 3 3 5 13 16 8** 32 24 ICB-3
No No 16 / 0 1 1 GB STI
- 1 STI
1C4 20 4 3 5 12 16 8** 32 24
ICB
No No 8/0 1 333 MB STI6
- 1 STI
1C5 20 5 3 5 11 16 8** 32 24
ISC-3, ICB-3,
1C6 20 6 3 5 10 16 8** 32 24 — — — — 32 channel maximum
ICB

1C7 20 7 3 5 9 16 8** 32 24 IC Channel8 No No 32 / 0 2 u-code support

1C8 20 8 3 5 8 16 8** 32 24 ISC-3, ICB-3,


ICB, — — — — 64 channel maximum
1C9 20 9 3 5 7 16 8** 32 24 IC Channels
microcode support -1
* up to ** up to 16 ICB links available via an RPQ if Compatibility I/O Cage is not
HiperSockets — — 4/0 1 CHPID for each
required. ***ICB Compatibility Links **** Up to 42 with RPQ 8P2233 defined HiperSockets
1. maximum 256 CHPIDS; 2. 28 I/O slots per nI/O Cage; 3. 22 I/O slots per cI/O Cage; 4. Supported and
I/O Slots required; 5. nI/O Cage plus cI/O Cage; 6. RPQ to 12: if no cI/O Cage present--RPQ to 16; 7. MM &
SM; 8. 2 defined channel
nI/O=new zSeries I/O Cage; cI/O=Compatibility I/O Cage

56 57
Summary of z900 Features

Processor Unit Assignments PCI Cryptographic Coprocessor


Models 101-109 Models 110-116, 1C1-1C9
Dual PCICC Cryptographic
PCICC
Up to 9 CPs Up to 16 CPs Features* Engines**
12 Up to 5 SAPs 20 Up to 8 SAPs
Up to 8 ICFs Up to 15 ICFs Minimum 0 0
PUs Up to 8 IFLs PUs Up to 15 IFLs
Maximum 8 16***
Processor Storage

z900 Increments 1 2
100-109 110-116 1C1-1C9
Models
* requires nI/O Cage;1 card slot/feature
** 2 Crypto engines/feature; 2 CHPIDs per feature
Minimum 5 GB 10 GB 10 GB *** Combined total of PCICC and PCICA engines
cannot exceed 16

Maximum 32 GB 64 GB 64 GB
PCI Cryptographic Accelerator
Storage sizes: 5, 6, 7, 8, 10, 12, 14, 16, 18, 20, 24, 28, 32, 40, Dual PCICA Cryptographic
PCICA
48, 56, 64 GB Features* Engines**

Channels Minimum 0 0
FICON/FICON
Parallel ESCON
Express Maximum 6 12***

Minimum 0 0 0
Increments 1 2
Maximum 88/96* 256 96 * requires nI/O Cage;1 card slot/feature
**2 Crypto engines/feature; 2 CHPIDs per feature
*** Combined total of PCICC and PCICA engines
Increments 3**/4 4 2 cannot exceed 16

*Available with G5/G6 upgrade or via RPQ for new builds.


**Available with G5/G6 upgrade only (existing 3-port parallel card).
Notes: OSA Networking Features
- Min. 1 Feature of Parallel, ESCON or FICON or FICON Express
required OSA-Express* OSA-2**
- Total CHPIDS all types (parallel, ESCON, FICON, FICON Express, Features Features
OSA, PCICC/CA, ISC-3, ICB, ICB-3, IC-3): Maximum 256
Minimum 0 0
Coupling Links

Links IC ICB-3 ICB ISC-3 Max # Links Maximum 12 12

32
z900 - 100 CF 32 16 16 42 64 Increments 1 1
w/RPQ
* GbE, FEN, ATM, TR; 2-ports/feature; 2 CHPIDs/feature;
8 requires nI/O Cage; one card slot/feature
z900 Server 32 16 12/16 32 32 ** Requires cI/O Cage: TR: 2-ports/feature, FDDI:
w/RPQ 1 port/feature; 1 CHPID/feature; 1 card slot/feature

58 59
z900 Frame and I/O Configuration Content: parallel channels, OSA-2 Token-Ring/FDDI ports and
Planning for I/O ESCON 4 port cards from the previous G5/G6 system
The following frame configurations and I/O tables show and one nI/O Cage to support the new I/O cards. Each
the capability and flexibility built into the I/O subsystem. cI/O Cage has 22 card slots (2 cages, 44 slots). The
The single frame system table (new build systems) ESCON 4 port cards are available on the z900 only
shows that up to 256 ESCON channels can be through an upgrade from a G5/G6. Any ESCON overflow
contained in one new zSeries I/O (nI/O) Cage. 256 out of the cI/O Cages to accomplish the upgrade and
ESCON channels, with the 16 port card (1 port is a any ESCON new channel requirements are provided by
spare), will fill 18 of the 28 card slots in the nI/O Cage the new ESCON 16-port card in the nI/O Cage. A typical
demonstrating the packaging efficiency of the current upgrade consisting of 200 ESCON channels (50 cards),
design. A more typical configuration of 200 ESCON 24 parallel channels (6 four port cards) and 2 FDDI
(14 cards) and 20 FICON (10 cards) channels, 4 ports (2 cards) would have an ESCON overflow of 56
OSA-Express ports (2 cards) and 8 ISC links (2 cards) still ESCON channels into the nI/O Cage. Fifty-six ESCON
fits within the single nI/O Cage (using all 28 card slots). (new) channels would require four of the twenty-eight
slots in the nI/O Cage. Customers upgrading from G5/G6
A customer planning to migrate to more FICON or to z900 should consider future channel requirements,
FICON Express channels or to a larger configuration, at FICON migration, and move to newer OSA-Express
a later date, might consider using the Plan Ahead Gigabit Ethernet, Fast Ethernet, ATM and Token Ring
feature to replace the single frame, single nI/O Cage adapters as part of the upgrade planning. The IBM
configuration with a two frame and two nI/O Cage Account team will help with this planning.
configuration. This action will allow nondisruptive
changes up to the limit of the two nI/O Cages. The
advantage of doing this is that: 1) the channel upgrade
will be completely nondisruptive and 2) the original I/O
A-Frame
will be distributed across the two nI/O Cages giving B
better balance and availability characteristics to the -
CEC F
upgraded system. Frame and cage configurations and r
their I/O capabilities are shown on the following pages. a New Build
m
nI/O Cage e 1 nI/O Cage
For a new build system with parallel channel and OSA-2
FDDI requirements, a Compatibility I/O (cI/O) Cage is
required. The single cI/O Cage has 22 card slots and Channel Type Features1 Maximum
can house up to 88 parallel channels (4 channels per ESCON (new) 3
18 cards max 256 channels
card) and up to 12 OSA-2 features. If requirements for FICON/FICON Express 16* 32 channels
greater than 22 card slots are required, a second cI/O OSA-Express4 12* 24 ports
Cage can be obtained through an RPQ; however, this PCICC 8* 16 engines
will reduce the number of nI/O Cages to a single cage in PCICA 6* 12 engines
the z900 A-Frame. Parallel, OSA-22 none none
* Maximum combined FICON, FICON Express, OSA-Express, PCICC/CA
Most upgrades from a G5/G6 system into a z900 system features is 16
will have two cI/O Cages in the Z-frame for packaging See notes on page 57.

60 61
z900 Frame and I/O Configuration Content:
Planning for I/O Z-Frame A-Frame
B
Z-Frame A-Frame -
New Build
B nI/O Cage CEC F
- r 2 nI/O Cages
CEC New Build a
nI/O Cage F 1 cI/O Cage
r 2 nI/O Cages m
a cI/O Cage nI/O Cage e
m
nI/O Cage e

Channel Type Features1 Maximum


Channel Type Features1 Maximum ESCON (new)3 18 cards max 256 channels
ESCON (new) 3
18 cards max 256 channels FICON/FICON Express 32* 32 channels
FICON/FICON Express 32* 64 channels OSA-Express4 12* 24 ports
OSA-Express4 12* 24 ports PCICC 8* 16 engines
PCICC 8* 16 engines PCICA 6* 12 engines
PCICA 6* 12 engines Parallel2 22 88 channels
Parallel, OSA-22 none none FDDI2 12 12 ports
* Maximum combined FICON, FICON Express, OSA-Express, PCICC/CA
* Maximum combined FICON, FICON Express, OSA-Express, PCICC/CA
features is 32
features is 32

Z-Frame A-Frame Z-Frame A-Frame


B
- New Build B G5/G6 Upgrade to z900
-
nI/O Cage CEC F 3 nI/O Cages cI/O Cage CEC F 1 nI/O Cage
r
a
r 2 cI/O Cages
a
nI/O Cage nI/O Cage
m
m (typical upgrade)
e
cI/O Cage nI/O Cage e

Channel Type Features1 Maximum Channel Type Features1 Maximum


ESCON (new)3 18 cards max 256 channels ESCON (new)3 <18 cards** <256 channels
FICON/FICON Express 48* 64 channels FICON/FICON Express 16* 32 channels
OSA-Express4 12* 24 ports OSA-Express4 12* 24 ports
PCICC 8* 16 engines
6* PCICC 8* 16 engines
PCICA 12 engines
Parallel, OSA-22 none none PCICA 6* 12 engines
* Maximum combined FICON, FICON Express, OSA-Express, PCICC/CA ESCON (4-port)** 44 176 channels
features is 48 Parallel2 24 4-port 96 channels
32 3-port
Z-Frame A-Frame
B New Build Token-Ring2 12 24 ports
CEC
-
1 nI/O Cage FDDI2 12 12 ports
F * Maximum combined FICON, FICON Express, OSA-Express, PCICC/CA
r 1 cI/O Cage features is 16
a
** For upgrades, the ESCON 16 port card is used for ESCON overflow out of the
m
cI/O Cage nI/O Cage e Compatibility I/O cage and for fulfilling new ESCON requirements.
Notes:
1) Up to this number. nI/O Cage 28 features (cards) maximum, cI/O Cage 22 features
maximum. (see notes 3 & 6)
1
Channel Type Features Maximum 2) cI/O Cage only.
3) ESCON Configured in 4 port increments. Up to 28 channels in 2 cards, 60 channels in
ESCON (new)3 18 cards max 256 channels 4 cards, 88 in 6 cards, 120 in 8 cards, 148 in 10 cards, 180 in 12 cards, 208 in 14 cards, 240
FICON/FICON Express 16* 32 channels in 16 cards, 256 in 18 cards.
4) OSA-Express can be Gbit Ethernet/ Fast Ethernet /Token Ring/ATM
OSA-Express4 12* 24 ports 5) Up to 256 channels for the system including ISC-3, ICB and IC links, and Crypto
PCICC 8* 16 engines PCICC/CA.
6) ISC links, although not in the tables, are packaged in the nI/O Cage. To 8 ports in 2
PCICA 6* 12 engines cards, to 12 ports in 3 cards, 16 ports in 4 cards, 20 in 5 cards, 24 in 6 cards, 28 in 7
Parallel2 22 88 channels cards, 32 in 8 cards.
7) One nI/O Cage is standard in the A-Frame on all general purpose z900 processors.
FDDI2 12 12 ports 8) Total number of PCICC and PCICA cards cannot exceed 8 with a maximum number
* Maximum combined FICON, FICON Express, OSA-Express, PCICC/CA of 6 for PCICA.
features is 16

62 63
Physical Characteristics Coupling Facility—CF Level of Support

Models 101-116 and 1C1-1C9 CF Level Function


Minimum Maximum 10 System-Managed CF Structure Duplexing
1 Frame System 2 Frame System
without with
B-Frame B-Frame MQSeries shared message queues
WLM Multi-System Enclave Support
Power Requirements 9
50.60 Hz, kVA
5.3 13.7 Intelligent Resource Director **
IC3/ISC3/ICB3 peer mode**
Heat output KBTU/hr 18.1 46.8

Air flow CFM 800 1990


Dynamic ICF expansion into shared
Air flow m *3/m in 22.2 55.2 8 ICF pool
Floor space
Systems-Managed Rebuild
– Sq meters 1.32 2.81
– Sq feet 14.2 30.3 Shared ICF partitions on server models
7
DB2 Delete Name optimization
Including service
clearance
– Sq meters 3.04 6.18 ICB & IC
6
– Sq feet 32.7 66.5 TPF support
Approximate weight
– kg 917 1866 DB2 cache structure duplexing
– lbs 2021 4113 5 DB2 castout performance improvement
Height Dynamic ICF expansion into shared CP pool*
– cm 200.4 200.4
– inches 79.81 79.81 Performance optimization for IMS & VSAM RLS
Dynamic CF Dispatching
1
Reduced height feature = 70.5 inches with packaging 4
Without covers = 73.3 inches Internal Coupling Facility*
IMS shared message queue extensions
Model 100 Coupling Facility 3 IMS shared message queue base
Minimum Maximum
1 Frame System 1 Frame System DB2 performance
without with VSAM RLS
B-Frame and B-Frame and 2
I/O Cage ISC Links 255 Connectors/1023 structures for IMS Batch
DL1 (non-BMP)
Power Requirements
4.0 5.0
50.60 Hz, kVA
Dynamic Alter support
Heat output KBTU/hr 13.7 17.1 1 CICS temporary storage queues
Air flow CFM 600 750 System logger
Air flow m *3/m in 16.6 20.8
Notes:
Floor space
– Sq meters 1.32 1.56
– G1 base can be upgraded to CF Level 4.
– Sq feet 14.2 16.8 – G2 base can be upgraded to CF Level 5.
– G3 and G4 can be upgraded to CF Level 8.
Including service
clearance – G5 base CF level code is CF Level 6.
– Sq meters 3.04 3.28 – G6 base CF level code is CF level 8.
– Sq feet 32.7 35.3 – Detailed information regarding CF Levels can be found in
Approximate weight “Coupling Facility Level (CFLEVEL) Considerations” at url
– kg 644 1189 ibm.com/s390/pso/cftable.html
– lbs 1450 2621

Height * G3, G4, G5, and G6 only


– cm 200.4 200.4 ** - z900 required
– inches 79.81 79.81
1 Reduced height feature = 70.5 inches with packaging Please note that although a particular back level machine may be
Without covers = 73.3 inches
updated to a more current CFCC level, NOT all the functions of
that CFCC level may be able to run on that hardware platform, i.e.,
G3/G4 can be upgraded to CFlevel 8 but it cannot use dynamic
ICF expansion into shared ICF pool.

64 65
Fiber-Optic Cabling and System Connectivity handle single mode to multi-mode infrastructure
In designing the z900 I/O subsystem, primary goals conversions. In addition, there are ESCON trunk (Fiber
included openness, industry standards (Fibre Channel, Quick Connect; harnesses and brackets) features
SANs, connectors, etc.), significant bandwidth and I/O available for customers desiring the efficiency and
packaging density to handle the new e-business convenience of trunking solutions.
workloads as well as the traditional transaction
processing environment. The connectivity alternatives IBM Global Services can supply the full complement of
available in this high performance environment have cabling solutions including the conversion cables and
significantly increased the number of cabling MCP kit cables plus jumper cables of various lengths,
alternatives available to connect systems to I/O and connector arrangements, cable types, and ratings for
networks. It is important that the customer's new connection requirements. IBM Global Services
infrastructure planning consider the technology also offers Fiber Transport Configuration Service; a
alternatives. In addition, fiber-optic trunking via Fiber specific cable service contract for the planning, supply
Quick Connect is available for ESCON on the z900. In and installation of up to 50 cables or a full service
planning for z900 systems, the customer should contract for implementation of a Fiber Transport
reference Planning for: S/390 Fiber Optic Links structured cabling system.
(ESCON, FICON, Coupling Links, and Open System
Adapters), GA23-0367, and the Installation Manual Bottom line: IBM offers a wide range of cables and
Physical Planning (IMPP) manual on Resource Link cabling solutions to meet a customer's requirements
(ibm.com/servers/resourcelink). for connecting to I/O and networks.

IBM offers fiber-optic cable features orderable with the


z900 and/or IBM Global Services to provide the cable
connectivity necessary for installing the z900. A limited
number of cable features are orderable with a z900
system which adapt the system to an existing cable
infrastructure. Included are two 2 meter conversion
cables which provide connector integration to the new
ESCON, ETR and ISC-3 I/O cards and five 2 meter
Mode Conditioning Jumper (MCP) kit features that

66 67
z/OS

While z900 servers are supported by a multitude of • A new installation and configuration infrastructure that
operating systems, their most advanced features are reduces the skills required to install and configure z/OS
powered by z/OS. z/OS is the foundation for the future of and related products.
zSeries, an integral part of the z/Architecture designed
• A new software pricing model designed to support
and developed to quickly respond to the demanding
quality of service requirements for enterprise e-business. e-business reality.
• System-managed CF Structure duplexing.
z/OS is a new, robust operating system that is based on
the new 64-bit z/Architecture. It delivers the highest
Intelligent Resource Director
qualities of service for enterprise transactions and data,
Intelligent Resource Director (IRD) is a new feature of
and extends these qualities to new applications using the the z/Architecture which extends the Workload
latest software technologies. It provides a highly secure,
Manager to work with PR/SM on z900 servers to
scalable, high-performance base on which to build and
dynamically manage resources across an LPAR cluster.
deploy Internet and Java®-enabled applications, providing An LPAR cluster is the subset of the systems that are
a comprehensive and diverse application execution
running as LPARs on the same CEC. Based on
environment. z/OS takes advantage of the latest software
business goals, WLM can adjust processor capacity,
technologies: new object-oriented programming models channel paths, and I/O requests across LPARs without
that permit the rapid design, development and deployment
human intervention.
of applications essential to enterprise e-business. It
protects your investment in your present S/390 IRD assigns resources to the application; the
applications by providing options for integrating existing application is not assigned to the resource. This
applications within your e-business infrastructure. It capability of a system to dynamically direct resources
provides a solid base for delivering on the benefits of to respond to the needs of individual components
industry-specific UNIX® and e-business applications, within the system is an evolutionary step. It enables the
supporting new technologies such as Enterprise Java system to continuously allocate resources for different
Beans™, XML, HTML, and Unicode. It supports such applications, and this helps to reduce the total cost of
technological advances as Parallel Sysplex processing, ownership of the system. IRD is made up of three parts
TCP/IP networking capability and complies with industry which work together to respond to the demands of
standards. e-business:
• LPAR CPU Management
z/OS helps make critical data and processing functions
accessible to end users regardless of their location in the • Dynamic Channel Path Management
heterogeneous e-business world. The z/OS base includes • Channel Subsystem Priority Queuing
z/OS Communications Server, which enables: world-class
TCP/IP and SNA networking support, including enterprise (see page 8)
class dependability, performance, and scalability; highly
secure connectivity; support for multiple protocols, and 64-bit Real Storage Support
efficient use of networking assets. z/OS provides 64-bit real storage support for the new
z900 servers. z/OS will continue to provide 31-bit real
The z/OS operating system combines many features and expanded storage support for G5/G6, and
that change the playing field of I/T infrastructure design: Multiprise 3000 servers. 64-bit real support will
• Intelligent Resource Director expands the capabilities eliminate expanded storage and help eliminate paging.
of z/OS Workload Manager to react instantly to chang- The 64-bit real support may allow you to consolidate
ing conditions and prioritize critical business workloads. your current systems into fewer LPARs or to a single
native image.
• Support for 64-bit real memory and initial support of
64-bit virtual storage.

68 69
These z/OS functions are enhanced to exploit 64-bit answers to high-level questions. After the configuration
real storage above 2 GB: parameters have been specified, msys for Setup can
• Traditional Access Methods (BSAM, QSAM, and others) automatically update the system configuration directly.
• VSAM for extended format data sets The user can see in detail what the changes will be
before they are made. Furthermore, msys for Setup
• Hierarchical File System (HFS)
introduces an LDAP-based z/OS management directory
• Extended Remote Copy (XRC) that will become the central repository for all
configuration data.
These IBM products are enhanced to exploit 64-bit real
storage above 2 GB: msys for Setup makes setting up a Parallel Sysplex
• DB2 Version 6 (with PTF) resource sharing environment much quicker and
• IMS Version 7 easier by defining and implementing the required:
• policies
IBM is working with many software vendors to enhance • parmlib specifications
64-bit exploitation. IBM and ISV software products that • security
run in 31-bit mode under OS/390 V2R6 through OS/390
V2R10 will run unchanged on z900 servers under z/OS V1R1 focuses on resource sharing configurations,
OS/390 V2R6 through OS/390 V2R10. including support for Intelligent Resource Director, XCF
signaling, and Global Resource Serialization star.
In addition, z/OS V1R2 delivers basic 64-bit virtual
storage management support. Assembler programs can z/OS V1R2 extended this support to cover additional
obtain virtual storage above 2 GB for storing and tasks, such as setting up a system logger environment
manipulating data. as well as customizing enhanced catalog sharing. Also
you can define basic IP settings, define network
Managed System Infrastructure for Setup devices and links or set up FTP or Telnet 3270 servers.
(msys for Setup) Finally, ISPF definitions can be generated through msys
z/OS starts a major ease-of-manageability initiative with for setup.
the introduction of Managed System Infrastructure for
Workload License Charges — Pay for what
Setup (msys for Setup) as a new base element. msys
you need
for Setup offers a new approach for installing and
Workload License Charges is IBM’s newest software
configuring z/OS and products running on z/OS that
pricing strategy designed to support e-business reality.
result in major productivity improvements.
With Workload Licence Charges, customers pay only
msys for Setup allows for the usage of consistent for the software capacity that they need. Because the
interfaces with wizard-like configuration dialogs. These new pricing structure is designed around the concept
new dialogs reduce the skill requirements for setting of paying for what you need, customers can grow a
up products, freeing up scarce and valuable personnel single workload at a time, with ease and granularity.
for other tasks.
Highlights
msys for Setup builds upon the Web-based wizard • New software pricing strategy designed to support
technology that has been introduced in a number of e-business reality
areas. msys for Setup employs the same easy, • Changes the metric of software pricing from machine-
interview style as the wizards for defining the capacity to sub-machine capacity
customization parameters. • Allows customers to grow one workload at a time, in a
The dialogs use defaults and best practices values simple and granular fashion
whenever possible and derive low-level values from • Provides enhanced price performance

70 71
• Enables customers to add excess hardware capacity Sysplex Distributor
without incurring additional software charges Introduced in OS/390 R10, Sypslex Distributor is a
• Permits customers to define and pay for software on software-only means of distributing IP workload across
their (four hour rolling) average requirement instead of a Parallel Sypslex cluster. Client connections appear to
the peak requirement be connected to a single IP address yet the
connections are routed to servers on different z900 or
System-Managed CF Structure Duplexing S/390 servers. In addition to load balancing, Sysplex
z/OS V1R2 introduces System-Managed Coupling Distributor simplifies the task of moving applications
Facility (CF) Structure Duplexing to significantly within a Parallel Sysplex environment.
enhance Parallel Sysplex availability. It provides a
robust failure recovery capability via CF structure System Services
redundancy, and it enhances Parallel Sysplex ease of z/OS Version 1 Release 2 base elements
use by reducing the complexity of CF structure Base Control Program (BCP)
recovery. These benefits are achieved by creating a JES2
duplicate (or duplexed) copy of a CF structure and ESCON Director Support
then maintaining the two structure instances in a MICR/OCR Support
synchronized state during normal CF operation. In the Bulk Data Transfer base
DFSMSdfp ™
event of a CF related failure (or even a planned outage EREP/MVS™ V3R5
of a CF), failover to the remaining copy of the duplexed High Level Assembler V1R4
structures is initiated and quickly completed ICKDSF R16
transparent to the CF structure user and without ISPF
TSO/E
manual intervention. 3270 PC File Transfer Program V1R11
TCP/IP Networking Enhancements FFST™/ESA
TIOC
z/OS can provide near continuous availability for TCP/IP IBM License Manager
applications and their users with two key features in
z/OS: Sysplex Distributor and VIPA nondisruptive z/OS Version 1 Release 2 optional priced features
Takeover. This is a prime example of IBM innovation DFSMSdss ™
and integration in software and hardware to bring value DFSMShsm ™
DFSMSrmm ™
added qualities — namely very high resiliency and JES3
availability — to the z900 and S/390 networking Bulk Data Transfer (BDT) File to File
environment. BDT, SNA, NJE

Virtual IP Address Nondisruptive Takeover


The backbone of the z/OS system is the Base Control
VIPA represents an IP address that is not tied to a
Program (BCP) with JES2 or JES3. These provide the
specific hardware adapter address. The benefit is that if
essential services that make z/OS the system of choice
an adapter fails, the IP protocol can find an alternate
when workloads must be processed reliably, securely,
path to the same software, be it the TCP/IP services on
with complete data integrity and without interruption.
z900 or an application.
Highlights:
VIPA Takeover introduced in OS/390 R8 supports Workload Manager (WLM) addresses the need for
movement to a backup IP stack on a different server in a managing mixed workload distribution, load balancing
Parallel Sypslex cluster in case of a failure of the and the distribution of computing resources to
primary IP stack. VIPA Nondisruptive Takeover competing workloads. It does this while providing fewer,
enhances the initial R8 functions, providing VIPA simpler system externals, performance management
takeback support. This allows the movement of goals expressed in Service Level Agreement terms,
workload back from the alternate to the primary IP stack. automatic work and resource management. All this is

72 73
done with a single policy that can be used across the managers have cleaned up for the failing system and
sysplex to provide a single control point, eliminating the can therefore restart registered JOBs/STCs as quickly
need to manage each individual image. as possible. In system failure cases, ARM uses policy
and Workload Manager (WLM) to balance the restart of
IB M L
Lii cens
censeeM ana
Mana
anagg e r, a new base element of z/OS, is failed elements across surviving systems.
not yet enabled for use.
Base Control Program
D F S M Sd fp provides a comprehensive set of functions
Sdfp
•The BCP has support for the IEEE floating point hard-
to manage storage resources on the system. Data
ware which is standard on all servers supported by z/OS.
management functions support storage and retrieval of
data on disk, optical and tape devices. Program •Allocation improvements allow DB2 to support custom-
management functions allow creation and retrieval of ers with a requirement for more than 10,000 DB2
executable program libraries. Device management dynamically-allocated table spaces.
functions provide the means to define and control the •Open Systems Adapter 2 (OSA-2) — support provides
operation of input and output storage devices. Fast Ethernet LAN attach communication.
Distributed File Manager (DFM) supports access to •Dump command parameters can be supplied via a
remote data and storage resources. PARMLIB member. Parallel Sysplex problem determina-
The existence of a completely integrated set of system tion provides convenient specification of multiaddress
services assures that a customer can focus on space and multisystem dumps.
extracting the maximum business value from the z/OS •Display Logger command improves Parallel Sysplex-
installation. The system manages the workload, program wide management of the System Logger. Logger ser-
libraries and I/O devices. Complexities are minimized viceability is improved with additional diagnostic and
and problem determination is facilitated with recovery data capture capability. DASD Logger Offload process-
and reporting facilities. ing is improved by operating asynchronously to overlap
operations and more efficiently utilize the I/O sub-
Parallel Sysplex system.
Dynamic Workload Balancing •IOS Channel Path Availability with asynchronous WTOR
Data sharing and workload balancing enable work to processing and enhanced recovery actions.
be directed to available processors “on the fly” and also
•Recoverable Resource Management Services (RRMS)
permit servers to be dynamically added to the cluster
recognizes Resource Managers that manage work
without requiring costly downtime. This can be
contexts as Work Managers. RRMS systems manage-
accomplished without splitting applications or
ment supports Work Manager names, wild card filtering,
databases across multiple servers, a time-consuming
sorted displays, and exit duration reports with the result
and expensive process, or requiring significant new
that it is easier to locate and manage RRMS coordi-
investments in staff or systems management.
nated transactions. Connections from any RRMS man-
Auto Alter aged application-enabled environment to access IMS
Dynamically tunes many of the CF structure sizes and DB databases is supported by IMS Version 6 Open
object ratios. Database Access (ODBA) using the IMS Database
Automatic Restart Manager (ARM) Resource Adapter (DRA).
To minimize the impact of a failed database manager JES2 and JES3
or a failed system, this database manager with its •In z/OS V1R2, JES2 and JES3 allow an installation to
associated transaction manager regions must be have up to 999,999 jobs. In addition, both provide the
restarted as soon as possible. Since ARM is part of installation the ability to obtain (spinoff) their JESlog
z/OS, it quickly knows of address space, JOB, and data sets prior to job completion.
system terminations. ARM knows when all resource

74 75
System Management Services The unique and rich functions of System Automation
for OS/390 (SA OS/390) Version 2.1 (separately
z/OS Version 1 Release 2 base elements
orderable) can ease z/OS and OS/390 management,
HCD
SMP/E
reduce costs, and increase application availability. SA
Managed System Infrastructure for Setup OS/390 automates I/O, processor, and system
Managed System Infrastructure for Operations operations, and includes “canned” automation for IMS,
CICS, Tivoli OPC, and DB2. Its focus is on Parallel
z/OS Version 1 Release 2 optional priced features
Sysplex automation, including multi- and single-
RMF
SDSF system configurations, and on integration with
HCM end-to-end Tivoli enterprise solutions.

With the new patented manager/agent design, it is


z/OS provides systems management functions and
now possible to automate applications distributed
features to manage not only host resources, but also
over a sysplex by virtually removing system
distributed systems resources. These capabilities have
boundaries for automation.
a long, successful history of usage by S/390 customers.
z/OS has enhanced many of these systems management Highlights:
functions and features to provide more robust control and •In z/OS V1R1, msys for Setup makes setting up a
automation of the basic processes of z/OS. Parallel Sysplex resource sharing environment much
quicker and easier by defining and implementing
z/OS introduces Managed System Infrastructure for required policies, parmlib specifications and security
Setup (msys for setup), a new base element, offering a settings
new approach for installing and configuring z/OS and
•z/OS V1R2 makes it easier to setup ISPF and IP Ser-
products running on z/OS. msys for Setup allows usage
vices, and provides increased benefits for setup of a
of consistent interfaces with wizard-like configuration
Base or Parallel Sysplex environment
dialogs. These new dialogs reduce the skill
requirements for setting up products, freeing up scarce •RMF can show response time distributions and work
and valuable personnel for other tasks. manager delay data, as well as reporting information
according to periods, even for report classes
z/OS Managed System Infrastructure for Operations •RMF can also show the contention for Cryptographic
(msys for Operations) in z/OS V1R2 makes well-proven Coprocessors, including a description of which
automation technology part of z/OS. It provides self- workloads are using or are delayed in access to the
healing for some critical system and sysplex resources cryptographic coprocessors
and simplifies the day-to-day operation of a z/OS
Parallel Sysplex environment. It monitors sypslex Benefits can include:
specific events to avoid single points of failure (such • Improve the productivity of system programmers
as no alternate Couple Data Set) or sysplex-wide • Reduce the skill level for configuring z/OS compo-
outages due to operator mistakes or system load. nents or products

SMP/E provides the ability to install software products


and service either from DASD or tape, or directly from a
network source, such as the Internet. By installing
directly from a network source, SMP/E is enabling a
more seamless integration of electronic software
delivery and installation.

76 77
Security Services RACF
•Digital Certificates can be automatically authenticated
z/OS Version 1 Release 2 base elements
without administrator action.
Cryptographic Services (Integrated Cryptographic Service
Facility, Open Cryptographic Services Facility, System SSL)
•Administrative enhancements enable definition of pro-
files granting partial authority. Handling of new pass-
z/OS Version 1 Release 2 optional priced features words and removal of class authority are simplified.
SecureWay® Security Server, which includes: •Network Qualified LU Names allow non-unique LU
- RACF® Names within interconnected networks.
- Network Authentication and Privacy Service2 (Kerberos services)
- LDAP Server2 • A protected user ID can be created to ensure user IDs
- Firewall Technologies assigned to UNIX, UNIX daemons, and other important
- DCE Security Server (OSF DCE level 1.2.2) started tasks and subsystems cannot be used for other
- Open Cryptographic Enhanced Plug-ins purposes. It will also protect the IDs from being revoked,
z/OS Version 1 Release 2 optional no charge features either accidentally or intentionally, with invalid pass-
Network Authentication Service Level 31 word attempts.
System SSL Security Level 31 •Identification and audit trail for a UNIX System user.
Open Cryptographic Services Facility Security Level 31 •Support for Tivoli management of all user segments.
1
Export considerations •Customers designing e-business applications need a
2
These functions are part of the base z/OS and do not require SecureWay Security way to associate more users under a RACF Group defi-
Server feature license
nition, so RACF allows you to create a new kind of
Group that can contain an unlimited number of users.
The optional SecureWay Security Server for z/OS
combines the traditional benefits of RACF with an SSL
integrated set of security functions essential to e- •Support for the Transport Layer Security (TLS) Version 1
business applications. It forms the basis for all security protocol (RFC 2246).
services from traditional applications, UNIX •Support for PKIX-compliant Certificate Revocations
applications, and distributed systems. Access to Lists (CRLs) created by the Tivoli SecureWay Public
resources can be selectively controlled, audited, and Key Infrastructure.
managed with appropriate centralized or decentralized •Ability for applications to create multiple SSL environ-
control as required by each installation. ments within a single process. An application can now
modify environment attributes without terminating any
Highlights: SSL sessions already underway.
•Integrated package of security and directory functions,
with advanced security capabilities LDAP
•Tivoli administration interface •Client access to information in multiple directories is
supported with the LDAP protocol.
Benefits can include: •LDAP Server now at V3 level. Bulk load utility for popu-
•Extensive security controls over emerging e-business lating the LDAP Directory.
opportunities •The LDAP server supports thousands of concurrent cli-
•Flexible control of access to protected resources, ents, increasing the maximum number of concurrently
including installation-defined items connected clients by an order of magnitude.
•Choice of centralized or decentralized control of secu- •The LDAP Server supports authentication using
rity profiles Kerberos credentials.
•Choice of platform for DCE security server, with RACF •An LDAP Configuration Utility will easily automate a
for access control, and remote administration basic setup.
• Support for the Kerberos V5 network authentication •An LDAP client can find an LDAP server with informa-
protocol tion in a Domain Name System (DNS) server without
knowing the LDAP server’s host name or IP address in
advance.

78 79
ICSF Application Enablement Services
Also featured as part of Security Management is the
z/OS Version 1 Release 2 base elements
Integrated Cryptographic Service Facility, which
Language Environment®
provides cryptographic functions for data security, data C/C++ IBM Open Class Library
integrity, personal identification, digital signatures, and DCE Application Support
the management of cryptographic keys. These ENCINA® Toolkit Executive
functions are provided through the combination of
z/OS Version 1 Release 2 optional priced features
secure cryptographic hardware, the ICSF cryptographic
C/C++ with Debug Tool
API, and the ICSF administration interface. C/C++ without Debug Tool
DFSORT™
z/OS supports the PCI Cryptographic Coprocessor GDDM®-PGF V2R1.31
feature for the S/390 Parallel Enterprise Server G5/G6 GDDM-REXX V3R21
and z900 servers. These pluggable cryptographic HLASM Toolkit V1R31
coprocessor cards provide additional cryptographic 1. z/OS V1R2 non-exclusive elements can be purchased standalone
capabilities, and more flexible and scalable crypto- z/OS provides a solid infrastructure in which you can
graphic function. ICSF is able to route cryptographic build new applications, extend existing applications,
requests to either the CMOS Cryptographic and run existing OLTP and batch processes.
Coprocessor or to a PCI Cryptographic Coprocessor
based on workload and capability of the coprocessors. Highlights:
These coprocessors therefore increase e-business The IBM ^ brand is about uncompromising
capacity, as well as provide an opportunity for more flexibility in selecting, building and deploying the
rapid implementation of new cryptographic algorithms. applications your business needs. Toward that end, IBM
offers the industry’s broadest range of platforms and
z/OS V1R2 provides leading edge SSL performance operating systems. And IBM is committed to industry-
through support of the z900 IBM PCI Cryptographic standard, cross-platform technologies — such as Java,
Accelerator (PCICA), a fast cryptographic processor XML, HTML, SOAP and UDDI — that are at the heart of a
designed to provide leading edge performance of the flexible e-business infrastructure. Support for these
complex RSA cryptographic operations used in the standards in our key middleware — including DB2
Secure Sockets Layer (SSL) protocol. This will benefit Universal Database, WebSphere Application Server
functions that use System SSL, such as the z/OS HTTP and MQSeries — means you won’t be locked in to a
Server (and WebSphere), TN3270 server, LDAP server single platform as your business grows. As a result, you
and CICS Transaction Gateway server. always have the flexibility to deploy applications in a
cost-effective way.
Network Authentication and Privacy Service
•A component of SecureWay Security Server, Network Today’s applications require much more than a set of
Authentication and Privacy Service, provides authenti- traditional development tools. To provide true business
cation, delegation and data confidentiality services value they must also:
which are interoperable with other industry implemen- •Enable rapid application development to deliver timely
tations based on the MIT Kerberos V5 reference imple- business function
mentation. Network Authentication and Privacy Service, •Be portable and interoperable
administered with RACF commands, supports both the •Enable function to be delivered with performance
native Kerberos API functions as well as the GSS-API across the network
Kerberos security mechanism and does not require DCE. •Allow access to data in a heterogeneous environment
•Be manageable and highly secure

80 81
Benefits can include: these function calls and improves the performance of
•Improved development productivity, cycle time and these applications. In order to exploit the bulk of “high
quality performance linkage” customers must recompile their C
•Reuse of existing code as objects through “wrappering” and C++ programs under the new XPLINK environment.
•Support for multiple language environments
•Improved end user productivity through provision of Network Communication Services
GUIs for existing applications z/OS Version 1 Release 2 base elements
•Choice of many high level languages enables use of z/OS Communications Server (Multiprotocol/HPR Services,
existing skills TCP/IP Services, SNA/APPN Services)
OSA Support Facility
•Improved interoperation and portability of code and
objects between heterogeneous platforms z/OS Version 1 Release 2 optional no-charge features
z/OS Communications Server Security Level 31
•Distributed applications development z/OS Communications Server Network Print Facility
Language Environment (LE) 1. Export considerations

•Performance improved by HEAPPOOLS, message filter-


ing, and pathlength reductions. The z/OS base includes z/OS Communications Server,
which enables: world class TCP/IP and SNA networking
•Serviceability changes adding CEEDUMP and IPCS
support, including enterprise class dependability,
support.
performance and scalability; highly secure
C/C++ connectivity; support for multiple protocols, and
•OS V1R2 provides two levels of the C/C++ compiler as efficient use of networking assets.
migration aid. One is identical to the compiler provided Highlights:
with OS/390 V2 R10. The other is fully compliant with •HiperSockets is a new function introduced in z/OS
the ISO C++ standard, also known as the ANSI C++ V1R2 for ^ zSeries which provides very high
standard. This includes support for ISO Standard C++ speed, low latency TCP/IP data communications across
Libraries, and other language features of ISO C++ 1998. LPARs resident within the same z900 server.
New application development involving C++ classes HiperSockets acts like a TCP/IP network within the
should make use of the C++ Standard Class Library, server.
instead of the C/C++ IBM Open Class Library.
• HiperSockets Accelerator, provides an “accelerated
•Enhanced ASCII support provides the ability to produce routing path” which concentrates traffic between
code that contains ASCII string literals and character OSA-Express external network connections and
constants. This allows ASCII dependent logic to con- HiperSockets connected LPARs. This function can
tinue working as on ASCII platforms, thus eliminating improve performance, simplify configuration, and
the need to find all such places in the code and con- increase scalability while lowering cost by reducing
verting them to EBCDIC when porting UNIX applica- the number of networking adapters and associated I/O
tions to z/OS. cage slots required for large numbers of virtual servers.
•The C/C++ Compiler includes support for IEEE Floating •The Sysplex Distributor has been extended to control
Point, and 64-bit long integer format. Support is also workload balancing, working in conjunction with Cisco
added to the C/C++ Runtime Library. MultiNode Load Balancing (MNLB).
•Performance and serviceability are improved, with a •Host-based Intrusion Detection Services (IDS) comple-
single optimization level, BCD format enhancements, ments network-based IDS sensors and scanners by
and allocation for stack storage. providing defense mechanisms that discard attacking
•Class library header in HFS. packets before they cause damage, discard packets
•Extra Performance Linkage (XPLINK) exceeding established thresholds, and limit the number
of connections from “greedy” users.
When you run a C or C++ application, there is over-
head associated with each function call. The more •OSA-Express Queued Direct Input/Output support
highly functionalized a program, the more overhead. •LPAR-to-LPAR communication sharing a single
XPLINK cuts down on the overhead associated with OSA-Express ethernet or Token Ring adapter
82 83
•Integration of SNA/APPN, TCP/IP and AnyNet® Further improvement to TCP/IP sysplex support to
•High performance TCP/IP stack for all applications increase usability, availability and performance. Sysplex
Sockets, XCF Dynamics, and System Symbolics
•High Performance Native Sockets (HPNS) for TCP/IP
continue to improve performance, provide
applications
nondisruptive growth and reduce definition effort.
•High Performance Data Transfer for SNA applications Dynamic Fast Response Web caching within the TCP/IP
•Multinode Persistent Sessions for SNA applications service stack will provide up to two times improvement
running in a Parallel Sysplex environment in performance. Service Policy Agent offers Priority
•SNA 3270, Sockets, and APPC application support Networking for TCP/IP. Internet Security with enhanced,
further improved Firewall technologies and inclusion of
•High Performance Routing enhanced to additional the latest security standards such as IPSec and
S/390 network configurations SNMPv3. Enhanced addressing for TN3270 Server
•Native ATM support for high speed networking reduces consumption of resources and increase
•Easy access to host applications from Java-enabled capacity.
Web browser •Enhanced Multipath Load Balancing.
•Greater performance and mobility for CICS Sockets •Support for dynamic updates to Service Policy.
•Higher availability in Parallel Sysplex environments with
Benefits can include:
Virtual Addressing Takeover (VIPA).
•Simplified deployment of client/server applications
•Support of Internet Key Exchange (IKE) protocol to au-
•Function for new e-business Internet and
tomatically create and distribute encryption keys for dy-
intranet applications
namic IP clients.
•Multivendor, multiplatform connectivity
•Addition of Secure Sockets Layer (SLL) authentication
•High performance, high availability, network choice to the TN3270e server, to protect against unauthorized
•SNA class of service over IP networks access to SNA applications from TCP/IP clients.
•World class TCP/IP services •Support of Triple DES session encryption for SNA
•Dramatic improvements in TCP/IP performance include applications
optimization of the TCP/IP stack, and inclusion of a •Service Policy Agent for IBM Communications Server
number of performance related capabilities such as: Enhancements
•UNIX Sendmail is supported including the POP3 UNIX System Services
server.
• TN3270 sessions use Secure Sockets Layer, with in- z/OS Version 1 Release 2 base elements
creased number of supported ports (255). z/OS UNIX System Services Application Services
• Multi-Node Persistent Sessions (MNPS) includes re-
covery for Dependent LU Requester (DLUR)-owned z/OS contains the UNIX Application Services (shell,
LUs. This leads to full Parallel Sysplex exploitation utilities and debugger) and the UNIX System Services
with native IP networks, including nondisruptive ses- (kernel and runtime environment). The Application
sion-switching. Services Shell and Utilities provides the standard
•Client/Server Affinity improvements to allow recovery command interface familiar to interactive UNIX users.
over sync points in multiphase commit giving higher z/OS includes all of the commands and utilities
availability for these applications. specified in the X/Open Company’s Single UNIX
• Simple Network Management Protocol (SNMP) Dis- Specification, also known as UNIX 95 or XPG4.2.
tributed Protocol Interface (DPI) instance level regis-
tration. The z/OS UNIX Services Debugger provides a set of
•X-Motif uses Data Link Libraries (DLL) to reduce commands that allow a C language program to be
disk space. debugged interactively. The command set is familiar to
•Virtual Private Network (VPN) easier to establish. many UNIX users. UNIX System Services provide the
world of open UNIX-based computing to the

84 85
z/OS operating system. With Language Environment, UNIX Parallel Environment
they support industry standards for C programming, •Enhanced to support MPI 1.2 specification for C/C++
shell and utilities, client/server applications, and the applications.
majority of the standards for thread management and •Support for Intelligent Miner data with the use of WLM
the X/Open Single UNIX Specification. The combination facilities.
of open computing and z/OS allows the transparent
exchange of data, easy portability of applications, USS
cross-network management of data and applications, •Superuser controls allow for selective assignment of
and the exploitation of traditional MVS system strengths security by specific responsibility.
in an open environment. •More control for message queues.
Highlights: •Make porting from other platforms easier with support of
•X/Open UNIX 95 Branded the Magic Number (#!), double-square-bracket condi-
•Permanent z/OS UNIX Kernel tional testing, and other functions.
•Integrated/converged z/OS UNIX Sockets •Parallel Sysplex support to share UNIX file systems
benefits Web server applications and others who ac-
•Web application and UNIX C program performance
cess the hierarchical file system. This support can
improvements
make your data and information that reside in the HFS
•Improved z/OS UNIX setup available to your customers at any time, no matter
•Web threading improvements where the applications are running in the Parallel
•Flexibility improvements — HFS files can be program Sysplex environment.
controlled or APF authorized
•Addition of new UNIX commands Distributed Computing Services
•Multiprocess/MultiUser Kernel Support z/OS Version 1 Release 2 Base elements
Network File System (NFS)
Benefits can include: DCE Base Services (OSF DCE Level 1.1)
•Development and execution of UNIX applications — Distributed File Service (OSF DCE level 1.2.2)
z/OS is a UNIX platform
•Portability of applications to and from other platforms •NFS acts as a file server to workstations, personal com-
puters, or other authorized systems in a TCP/IP network.
•Use of UNIX development skills in a z/OS environment It also provides a z/OS client. The remote data sets or
•Consolidate multiple UNIX systems files are mounted from the mainframe to appear as lo-
•Scalability for high growth UNIX applications cal directories and files on the client system. NFS also
provides access to the Hierarchical File System (HFS).
UNIX
•DCE enables data encryption using the data encryp-
•Performance enhancements include recompiled and
tion standard (DES) algorithm and the commercial data
optimized functions within the kernel, and shell and
masking facility (CDMF).
utilities, addition of Socket Functions; use of Communi-
cation Storage Management buffer transfer instead of Highlights:
data movement; and optimized NFS Logical File System. • z/OS R2 introduces the zSeries File System (zFS). zFS
•Multiprocess/MultiUser allows faster process creation is a UNIX file system that can be accessed with the
for customers and reduced storage usage for servers. z/OS hierarchical file system file Application Program-
ming Interfaces (APIs). It will provide significant perfor-
•Semaphores without contention using the hardware
mance gains in most customer environments requiring
Perform Locked Operation (PLO) instruction.
files 8K in size or greater.
•Shared memory (captured storage) reduces real stor-
•Remote Procedure Call (RPC) lets calls between pro-
age when sharing large amounts of virtual storage.
grams running on different platforms appear as local
•UNIX System Services and UNIX debugger add support procedure calls.
for IEEE-floating point.

86 87
•Directory Services allows resources to be found any- e-business Services
where in an enterprise without the need to know local
z/OS Version 1 Release 2 Base elements
names.
IBM HTTP Server
•Security Services helps solve security problems com- Text Search
mon in a distributed environment by handling identifi-
z/OS Version 1 Release 2 Optional no charge features
cation and certification of users, clients, servers and
systems. IBM HTTP Server North America Secure

•Distributed Time Services synchronizes clocks running Businesses increasingly use the Internet to market
on different nodes products and conduct business with suppliers and
•DFS allows access to both z/OS UNIX Services customers. WebSphere Application Server for OS/390
Hierarchical File System and the OSF’s DCE Local File (separately orderable) enables the use of z/OS as a
System. Web server with the benefits of security, the utilization of
•DCE DFS brings the strengths of z/OS systems man- large storage capacity, centralized skills, a single point
agement software and high-speed unique file naming of entry and control, consolidation of multiple Web sites,
across enterprises. and secure Internet transactions.
•Client support for record-level access to PS, PDS, PDS/E Text Search is a powerful, full-text indexing and search
and VSAM data sets. server. It supports high-speed searching of Web sites, as
•Security based on Kerberos authentication with access well as other documents stored on the z/OS server.
to RACF data as well. Free text searching is supported, as well as Boolean
logic. Text Search returns a ranked list of hits. A full API
Benefits can include: set is provided so that Text Search functions can be
•Transparency of data and logic used in other programs and products. It can also be
•Distributed, consistent directory service used for Web-usage mining.

•Security for both clients and servers, including encryp- The IBM HTTP Server offers HTTP 1.1 compliance,
tion if required support for Java® technology, and the ability to manage
•Scalability of distributed applications Internet processing through the Workload Manager
(WLM). In addition, it provides:
•Interoperability and portability
•Automatic browser detection allows the server to re-
DFS spond to requests with the version of a Web page or
•Windows networking compatible file and print serving document appropriate for that browser.
is available with Native SMB File and Print Serving for
•Page counter and date/time information can be dis-
Windows Clients. SMB file serving enables z/OS to
played on a page as graphical images.
share HFS files with Windows workstations. This allows
development of applications on Windows, for deployment •Dramatic improvements in single and multiple proces-
on z/OS. z/OS also supports printing of SMB files without sor environments coupled with significant CPU require-
requiring that code be installed on the clients and without ment reductions provide higher throughput and shorter
requiring unique printer setup on the workstations. OS/2® response time.
client file serving support is also available. •Platform for Internet Content Selection (PICS) provides
a way for users to filter material they encounter on the
Internet and accept or reject the material based on its
rating.
•Web usage statistics.

88 89
Benefits can include: • IP PrintWay provides support for printers attached to the
•e-business with security network using TCP/IP, VTAM-controlled coax printers,
•Utilization of large storage capacity and for printers and servers over the internet using the
industry standard Internet Printing Protocol (IPP). Easy
•Single point of entry and control to use ISPF menus also enable management of distrib-
•Consolidation of multiple Web sites uted printers.
•Secure Internet transactions • The Print Interface supports print submission from ap-
plications running in UNIX System Services (USS), from
•Exploitation of z/OS WLM
Windows users via native Windows SMB, from applica-
tions on other servers, and over the internet using IPP.
Print Services • Data stream transforms let you print AFP applications
z/OS Version 1 Release 2 optional priced on printers using PCL, PostScript or PDF. You can also
Infoprint Server print PCL, PostScript and PDF output on AFP printers.
IP Printway™ • A transform from SAP to AFP and a certified SAP Output
NetSpool ™ Management System lets you print SAP application out-
z/OS Print Interface put on your fast, reliable AFP printers, and receive print
completion notification back at the SAP Application
Today, businesses are looking for better ways to get Server.
information to knowledge workers throughout the
enterprise, and to suppliers, partners and customers Benefits of consolidating your enterprise printing onto
around the globe. Application output that was once z/OS using Infoprint Server can include:
printed on centralized data center printers and mailed • Reduced total cost of ownership for distributed print
must now be electronically distributed and printed, or operations
presented over the Web. Infoprint Server provides a • Improved productivity with simplified print operations
reliable, high-availability, secure and scalable and management
foundation for your e-business output infrastructure. • Investment protection and leverage for your AFP appli-
Infoprint Server and its companion product, Infoprint cations and printers
Server Transforms, include a print interface, printer • Faster deployment of e-business initiatives with flexible
inventory, VTAM® application output capture program, output delivery options
and print drivers and management tools that let you
Test
manage any print job to any printer defined to Infoprint
To further reduce migration time, the z/OS product is
Server. These capabilities give you the flexibility to
system integration tested using a production-like
deliver the output of e-business anywhere you need it:
environment. This environment includes subsystems,
such as CICS, IMS and DB2. This additional testing
• Legacy CICS and IMS applications that generate SNA supplements existing functional tests, with a focus on
Character String (SCS) or 3270 output formats can print tasks performed by customers in the production
to LAN-attached PCL printers, without changes to the environment, thus helping establishments move more
application program. quickly to new function.
• Support for the output of e-business includes the abil-
ity to send output as e-mail instead of, or in addition to Publications
print. For a list of the publications available for z/OS, visit the
• A consolidated printer inventory lets you define all z/OS library Web site.
printers used with Infoprint® Server, and printers driven
by Print Services Facility (PSF) in one place. Printers
can be defined and modified from a single easy to use
interface.

90 91
Installation Considerations Table: Coexistence Supported Releases
OS/390 V2 R10 provides a strong foundation for z/OS. (OS/390 & z/OS)
You can install OS/390 V2R10 now and upgrade to z/OS Release Coexistence Release
V1R1 with the installation of the special z/OS V1R1 R6 R6, R5, R4, R3, R2 (1)
Upgrade Package for OS/390 V2R10. This product R7 R7, R6, R5, R4
upgrade package (PUP) will provide an easy path to R8 R8, R7, R6, R5
z/OS V1R1, and will install like a monthly RSU service
R9 R9, R8, R7, R6
upgrade. Non-OS/390 V2R10 customers, or customers
R10 R10, R9, R8, R7, R6 (2)
who would like to install z/OS V1R2, can order z/OS via
Customized Offerings (ServerPac, Systempac® and z/OS R1 z/OS R1 OS/390 R10 (3), R9, R8, R7, R6( 2)
CBPDO) for a complete system. z/OS R2 z/OS R2, z/OS R1, OS/390 R10 (3), R9, R8
Notes:
Customers are strongly encouraged to implement WLM 1. Coexistence of OS/390 R2 with OS/390 R6 supported as a special provision
2. Coexistence of OS/390 R6 with OS/390 R10 supported as a special provision
goal mode. WLM in goal mode continues to expand its
3. OS/390 R10 and z/OS R1 treated as one coexistence level
role in providing enhanced system efficiencies and
workload balancing on the S/390 and zSeries platforms.
This information is being provided to you to facilitate
Goal mode is required for much of the Intelligent
release planning and to ensure appropriate positioning
Resource Director functions. z/OS V1R2 will be the last
for future software migrations. Since each release can
release to support WLM compatibility mode.
normally be ordered for only a six month window, it is
very important that you order the required releases
z/OS V1R2 is supported on the following IBM
while they are available.
servers:
• S/390 Parallel Enterprise Server G5 and G6 (or compat- Since z/OS allows the JES element to be separately
ible server) (31-bit mode) staged, z/OS also supports the coexistence of certain
• Multiprise 3000 Enterprise Server (or compatible lower-level JES releases with the JES release provided
server) (31-bit mode) with z/OS.
• IBM ^ zSeries 900 servers (or compatible serv-
For additional information on the general z/OS policy, in
ers) (64-bit mode)
addition to the specific JES releases that can coexist
SOD: z/OS V1R3 and z/OS V1R4, expected to be available in 2H2002,
will run on G5, G6, MP3000 and z900 Servers (or compatible servers) with z/OS Release 2 JES in a multisystem configuration,
(31-bit mode). refer to z/OS Planning for Installation.
Note: Specific functions might only be available on the up-
Coexistence level systems, or it might be necessary to up-level all sys-
tems to enable some functions.
z/OS continues to give customers optimum
compatibility and flexibility as they migrate systems in a
multisystem configuration by allowing up to four
consecutive releases to coexist. Coexistence
considerations apply to multisystem configurations in
which there is resource sharing. This includes
non-Parallel Sysplex and Parallel Sysplex multisystem
configurations.

The following table describes the effect of this special


provision. The release shown in column 1 is the highest
release running in a multisystem configuration.

92 93
z/VM

z/VM is the follow-on product for VM/ESA customers. It z/VM also reduces storage constraints by eliminating
opens new and exciting opportunities for customers the 2 GB central storage limitation providing plenty of
who have built enterprise-wide automation and headroom for increasing e-business demands and
infrastructure enhancements on the VM platform in growing back-office applications within a single
support of their applications, database systems, and e- machine image. Customers experiencing real memory
business solutions. constraints can experience relief by running z/VM in
Using virtualization technology as a foundation, z/VM 64-bit mode on a zSeries server. This constraint relief is
offers new function and technology for customers who provided for ESA/390 guest operating systems. Support
want to exploit IBM Virtual Machine (VM) capabilities on for large real memory with z/VM may benefit customers
the mainframe. Virtualization technology allows running a large number of Linux for zSeries and S/390
customers to virtualize processor, communications, guest systems.
storage and I/O resources thus avoiding the overhead Additional enhancements in z/VM Version 3 include:
of planning, purchasing and installing new hardware to • Native FlashCopy for Enterprise Storage Server™ (ESS)
support new workloads. for high speed data copy
z/VM supports a wide variety of existing IBM zSeries and • Guest support enhancements for 3494 VTS and
S/390 servers. This offers today’s VM/ESA customer the FICON™-attached 3590 A60 Tape Controller
ability to maintain their existing VM environments on the Connectivity enhancements for TCP/IP Feature for z/VM:
latest technology while positioning themselves to grow • Improved security with the inclusion of a Secure
into the zSeries. Socket Layer (SSL) server
z/VM offers an ideal platform for consolidating Linux • Transparent data access to remote systems data with
workloads onto a single zSeries 900 or S/390 server. an NFS Client
Running Linux as a guest of z/VM enables you to run • Capability and usability improvements to FTP server for
hundreds of Linux images while benefiting from the Web browsers
reliability, scalability, availability, and serviceability • Reduced load on hosts with support for IP Multicasting
characteristics of zSeries 900 or S/390 servers. • Improved data transfer performance with QDIO sup-
porting Gigabit Ethernet, Fast Ethernet, and 155 ATM
z/VM Version 3 (Ethernet Lan Emulation)
z/VM offers a new technology base for customers • Support for the DFSMS/MVS® Program Management
looking to use IBM Virtual Machine technology on the binder and loader functionality to enhance application
industry’s best of breed server platform, zSeries 900. affinity between CMS and OS/390 or z/OS
When z/VM is running on a zSeries server, it’s possible
to run 64-bit capable OS/390 Release 10, z/OS, and z/VM Version 4
Linux for zSeries as guest systems of z/VM, in addition to Licensed using the IBM International Program License
ESA/390 guest operating systems such as OS/390, Agreement (IPLA) which includes a one-time charge
VSE/ESA, TPF, and Linux for S/390. To operate z/OS as a (OTC) pricing structure, z/VM Version 4 provides a great
guest of z/VM on a zSeries server, z/VM and z/OS must opportunity for customers to exploit zSeries 900 or
be operating in 64-bit mode. z/VM will allow customers S/390 servers at a substantially lower cost. z/VM Version
to develop and test their 64-bit Parallel Sysplex 4 is priced on a per-engine basis and is designed to
applications in a guest environment before putting them support the IBM Integrated Facility for Linux (IFL)
into production. This reduces the need to invest in processor features for Linux-based workloads and
separate standalone configurations. standard engines for all other zSeries 900 and S/390

94 95
workloads (an engine is also referred to as a central z/VM V4R2 extends its hardware support by exploiting
processor). z/VM Version 4 supports the IBM zSeries technological enhancements for the IBM ^
900, S/390 Parallel Enterprise Server™ Generation 5 and zSeries 900 (z900) server, such as:
6, and the S/390 Multiprise 3000, or equivalent. • faster communications between and among virtual
machines and logical partitions with HiperSockets
To help improve the performance of I/O to network • high-speed OSA-Express Token-Ring
devices, the z/VM Control Program (CP) provides a fast • improved z/OS Parallel Sysplex support with virtualized
CCW translator in z/VM V4R1 to translate a wide range Coupling Facility (CF) duplexing
of channel programs that perform I/O to network • FICON channel-to-channel adapter (CTCA) communi-
adapters. This fast translator for network devices cations for guests
includes support for 64-bit indirect data address words
(IDAWs), which allows guest machines to read and Connectivity enhancements to TCP/IP for z/VM, Level
write data above the 2 GB limit using 64-bit I/O 420 are designed to include a new Internet Message
addressing. In z/VM V4R2, 64-bit IDAW support was Access Protocol (IMAP) server, improved security of the
further extended to the existing DASD fast CCW TCP/IP stack by preventing some Denial of Service
translator. This enhancement is also designed to (DOS) attacks, and support for HiperSockets. A “Guest
include 64-bit IDAW support for DASD channel LAN” capability is designed to allow a VM guest to
programs that are simulated via the minidisk cache install a virtual HiperSockets adapter and connect it
(MDC). Additional guest-support improvements in z/VM with other virtual network adapters on the same VM
V4R2 that may benefit Linux virtual machine host system to form an emulated LAN segment.
performance include DASD fast CCW translation using
64-bit IDAWs and (when corresponding function is To aid application development within the Language
available from Linux for zSeries) enhanced page-fault Environment (LE), z/VM V4R2 consolidates existing VM
handling which allows a Linux guest to continue C-language sockets APIs into a single consistent
processing while the page faults are handled by z/VM. socket library.

When corresponding function is available from Linux RealTime Monitor (RTM) and Performance Reporting
for zSeries, z/VM V4R2 is also designed to provide the Facility (PRF) have been enhanced to support z/VM,
capability for Linux guest virtual machines to use clear- including new and changed monitor records in z/VM
key RSA functions of the IBM PCI Cryptographic Version 3. RTM simplifies performance analysis and the
Coprocessor (PCICC) or the IBM PCI Cryptographic installation management of VM environments. PRF uses
Accelerator (PCICA) for Secure Socket Layer (SSL) system monitor data to analyze system performance
acceleration on the zSeries 900 S/390 servers. z/VM and to detect and diagnose performance problems.
V4R2 offers some ease-of-use functions to manage RTM and PRF are priced, optional features of z/VM
Linux images and a migration facility to help existing Version 4 as is the Directory Maintenance Facility
customers of the Virtual Image Facility move their (DirMaint).
configuration files and data to a z/VM V4R2 system.
For further information see the z/VM and VM/ESA
Reference Guide, G320-9997-17.

96 97
Notes To learn more

Visit the S/390 World Wide Web site at


ibm.com/eserver/zseries or call IBM DIRECT at 1 800
IBM-CALL in the U.S. and Canada.

Australia 132 426


Austria 0660.5109
Belgium 02-225.33.33
Brazil 0800-111426
China (20) 8755 3828
France 0800-03-03-03
Germany 01803-313233
Hong Kong (20) 2825 6222
Hungary 165-4422
India (80) 526 9050
Indonesia (21) 252 1222
Ireland 1-850-205-205
Israel 03-6978111
Italy 167-017001
Japan 0120 300 426
Korea (02) 781 7800
Malaysia (03) 717 7890
Mexico 91-800-00316
Netherlands 020-513.5151
New Zealand 0800-801-800
Philippines (02) 819 2426
Poland (022) 878-6777
Singapore 1800 320 1975
South Africa 0800-130130
Spain 900-100400
Sweden 020-220222
Switzerland 0800 55 12 25
Taiwan (06) 2725 9300
Thailand (02) 273 4444
Vietnam Hanoi (04) 843 6675
Vietnam HCM (08) 829 8342
United Kingdom 0990-390390

98 99

You might also like