Contents
Base Parallel Sysplex support
z/OS V1R4 provides the following base Parallel Sysplex support:
Balancing your business workload
z/OS V1R4 improves the balancing of Workload Manager (WLM) managed batch
initiators between systems of a sysplex. On highly utilized systems, the number
of initiators can be reduced while more are started on low utilized systems.
This enhancement can improve sysplex performance with better use of the
processing capability of each system.
System-managed CF structure duplexing
A set of architectural extensions to
Parallel Sysplex is introduced with System-Managed CF structure duplexing
support for coupling facility structures. All three structure types (cache
structures, list structures, and locking structures) can be duplexed using this
architecture and provide improved availability. The new set of features is
quite extensive and, in total, represents a very significant extension to the
Parallel Sysplex Architecture.
Additional details on the availability and requirements for using
system-managed CF structure duplexing are available in
Software Announcement 102-181 (June 25, 2002).
In addition, the IBM technical paper
System-Managed CF Structure Duplexing (GM13-0103)
includes information about:
- The cost and benefits of using system-managed CF structure duplexing
- Determining which structures should be duplexed in a specific Parallel Sysplex
Reliability, availability, and serviceability (RAS) enhancements
Unauthorized CTRACE Writer: For unauthorized users of CTRACE, z/OS now offers
a writer process that will asynchronously copy trace records from in-storage
trace buffers to an external data set. This is similar to the External Writer
currently supported by CTRACE. This function is important to Java users. Now
Java can build its JVM trace using CTRACE for non-CICS environments (for
example, DB2, CB). Because Java trace records are typically written at a high
rate, and the JVM process could terminate as soon as it's done, it is
necessary, via Unauthorized CTRACE writer, to "harden" the trace records to an
external medium- a data set.
JES2 and JES3 enhancements: JES2 and JES3 have made the following RAS
improvements:
- JES3 — Refresh MAINPROC: Typically, in a JES3 environment, adding a system
to a sysplex has been a challenge because a warm start is required in order to
add systems, and that requires a sysplex-wide IPL. Refresh MAINPROC enables
you to add, delete, or change a MAINPROC statement (which is how JES3 defines a
member of a sysplex) during hot start with refresh. The major benefit of this
function is that, in most cases, it allows you to add systems to your sysplex
without having to perform a sysplex-wide IPL.
-
Redesigned interfaces between user address spaces and the JES2 address space
remove WAIT/POST bottlenecks at End of Memory and End of Job, which results in
improved processing of large dumps.
-
To reduce multisystem outages due to checkpointed data structure errors, z/OS
has broadened the scope of the verification and detection code that is used in
JES2 processing to detect and recover these kinds of errors.
-
Other improvements allow for JES3 startup despite errors that might prevent it
from initializing and also allow the customer to fix the error with relative
ease.
-
Further improvements to JES2 Resource Monitor enhance JES2's ability to
identify and report situations to which customers can react and respond without
having to resort to an IPL.
z/OS Managed System Infrastructure for Setup
Introduced with z/OS V1.1, msys for Setup is part of IBM autonomic computing
initiative and helps customers to enable functions as quickly and easily as possible.
The value of msys for Setup
is extended in z/OS V1R4 with the delivery of further msys for Setup
exploitation and major enhancements to the framework. The msys for Setup
framework improvements include:
- Multiple user support: Multiple User Support allows more than one user at a
time to work with msys for Setup on the same configuration. Furthermore, as
part of the user enrollment process, the msys for Setup user administrator can
control which msys for Setup workplace functions a user can access.
-
Improved user interface: The graphical user interface (msys for Setup
workplace) has been redesigned and is now easier to learn and use.
-
National language support: msys for Setup will provide messages, panels, and
help text translated into Japanese.
-
Job progress indication: The msys for Setup user will be informed about job
progress, which is very helpful especially during long jobs.
In addition to these framework highlights, the LDAP plug-in will provide the
ability to configure multiple LDAP servers. Also, configuration parameters will
be discovered automatically for servers which are established.
z/OS Managed System Infrastructure for Operations
Managed System Infrastructure for Operations (msys for Operations) is part of the
IBM
advanced self-healing and self-managing technology.
Introduced in z/OS V1.2, and further extended in z/OS V1.3,
attention is refocused in z/OS V1R4 because of
the important role msys for Operations plays in helping you to avoid outages.
msys for Operations functions control and manage both hardware and software resources;
thus, fully automated solutions may be possible.
msys for Operations simplifies
Parallel Sysplex and z/OS operator interaction, thereby detecting failure
situations and reacting to them quickly and precisely.
These factors are
essential to z/OS availability and directly affect the performance and
availability of many business applications.
z/OS customers can improve their operations and system availability by using
either:
- Panel-driven operator dialogs that assist in the control of couple data sets,
coupling facilities, and managing coupling facility structures
-
Background recovery routines that guard against console buffer shortages, long
running enqueues, or auxiliary storages
Several enhancements to msys for Operations include hardware policy validation,
critical ENQ resources detection, and the ability to remove failed systems from
a Parallel Sysplex. These enhancements build on msys for Operations' ability to
help improve the performance, recovery, and availability of business-critical
applications.
- Hardware policy validation: New function for msys for Operations will enable
cross-validation of the hardware configuration that is mapped out in the msys
for Operations policy against the actual running hardware configuration.
Validation of this type is critical to help accurately control LPARs and
determine that LPARS defined in the policy are actually on the correct system
and located in the correct partition by name and number. The validation takes
place during msys for Operations initialization and whenever disruptive
requests are about to be executed.
-
Hung command recovery: msys for Operations will help prevent multisystem
outages by detecting critical ENQ resources that have been held for an extended
period of time.
-
Removing failed systems from a Parallel Sysplex: Automated response to the
IXC402D message isolates and removes failed systems from a Parallel Sysplex as
quickly as possible. This can provide faster mean-time-to-recovery (MTTR) for
those system images the enterprise wishes to restart immediately should an
unavoidable outage occur.
For more information, visit the
msys for Operations Web site.
IBM Health Checker for z/OS and Sysplex
IBM Health Checker for z/OS and Sysplex helps improve availability by
reporting on active z/OS and Parallel Sysplex settings that are different from
best practices recommended by IBM or customer-defined settings. msys for
Operations will be enhanced to include an interface to these functions with
panels to simplify viewing of results.
You can download the stand-alone version of the IBM Health Checker for z/OS and
Sysplex tool from the
z/OS downloads Web page.
z/OS V1R4 Consoles Enhancements feature
As discussed in
Software Announcement 203-131 (May 13, 2003),
the z/OS V1R4 Consoles Enhancements feature is the first phase of IBM strategy to
enhance the operator messaging architecture of z/OS. The overall objective is
to improve system availability by enhancing the capacity and reliability of
message delivery. This capability was an optional feature in z/OS V1R4; it is
integrated into z/OS V1.5.
z/OS Communications Server
Sysplex-wide security association support:
SWSA allows IPSec protected workloads to realize all the benefits derived from
workload balancing, such as optimal routing of new work to the target system
and server application based on QoS and WLM advice, increased availability by
routing around failed components and increased flexibility in adding additional
workload in a nondisruptive manner. The key element that enables this solution
is the ability to distribute the IPSec processing including cryptography for a
single IPSec Security Association among systems in a sysplex environment.
SWSA also allows workloads with IPSec-protected traffic to exploit the benefits
of the VIPA Takeover function. With this enhancement, IPSec-protected
workloads can now be associated with Virtual IP Addresses (VIPA) that can be
recovered by other systems or servers in the case of a failure of a system
and/or server. This is accomplished by allowing IPSec Security Associations
(SA) to be automatically restarted on another processor in the sysplex when a
VIPA takeover occurs.
Sysplex-wide dynamic source VIPAs:
This provides the same single IP-address appearance for application instances
initiating outbound connections within a sysplex as Sysplex Distributor
provides for inbound applications. This capability simplifies IP address and
workload management, improves security, and allows greater scalability and
flexibility. For clients outside a Parallel Sysplex environment, Sysplex
Distributor provides a single-IP address appearance to application instances
spread across the sysplex. It also distributes incoming work among the various
instances.
TCP/IP support for Simple Network Time Protocol (SNTP):
z/OS V1.4 Communications Server includes SNTPD, a new TCP/IP daemon. By
providing a Simple Network Time Protocol (SNTP) Server, SNTPD enables
synchronization of time between a client and a server with the zSeries Sysplex
Timer. The availability of this function answers the need of customers who
have requested the ability to synchronize clocks between various platforms in
their network. TIMED is a TCP/IP daemon used to provide the time. TIMED gives
the time in seconds since midnight 1 January 1900. SNTPD is a new TCP/IP daemon
that is also used to provide the time in order to synchronize a network of
(S)NTP clients. Simple network time protocol provides for a more accurate time.
SNTPD does not replace TIMED but it is the preferred server for synchronizing
time in the network.
Enhanced dial processing for Enterprise Extender connections:
z/OS V1R4 enhances the dial processing for Enterprise Extender connections
to attempt automatic redial both in the case where an initial dial fails and in
the case where an existing connection fails. Prior to z/OS V1R4 Communications
Server, there was no mechanism to automatically attempt to redial a switched
Physical Unit for Enterprise Extender when a dial attempt failed or when an
existing connection INOPed.
High Performance Routing (HPR) exploitation by Enterprise Extender:
Prior to z/OS V1R4 Communications Server, you could not configure
Interchange Nodes (ICNs)
with links to some types of connection networks (such as ATM and Enterprise
Extender connection networks) due to a configuration restriction that did not
allow ICNs to exploit high performance routing (HPR) over connection networks
for sessions that cross from APPN into subarea.
(ICNs could compute session paths through these connection
networks for other APPN NNs or ENs that have links to them. However, the ICNs
themselves could not activate a link to these types of connection networks.
Instead, ICNs were required to predefine links to all other nodes on the
connection network, or allow APPN to compute session paths that include
additional nodes.)
z/OS V1R4 Communications Server eliminates this restriction for
Enterprise Extender connection networks.
In addition, this function will also allow HPR to be used (instead of
ISR) over other types of connection networks (like token-ring) for sessions
that cross from APPN into subarea.
Data management
DFSMStvs — Data access and business continuance support:
z/OS V1 DFSMS Transactional VSAM Services (DFSMStvs) is a new priced,
optionally orderable z/OS feature that enables batch jobs and CICS online
transactions to update shared VSAM data sets concurrently. Multiple batch jobs
and online transactions can be run against the same VSAM data sets, and
DFSMStvs helps ensure data integrity for concurrent batch updates while CICS
ensures it for online updates.
Prior to DFSMStvs, record-level sharing (RLS) with integrity was available to
CICS applications with VSAM RLS while transactional support for commit and
rollback was provided by CICS TS. With DFSMStvs, RLS with integrity and
transactional support for commit and rollback are now available to all non-CICS
VSAM applications.
DFSMStvs is designed to offer the following benefits:
- Contributes to the reduction or elimination of your batch window for CICS
applications and other VSAM applications by allowing concurrent batch and
online updating of VSAM recoverable data sets
-
Delivers increased system availability with simpler, faster, and more reliable
recovery operations for VSAM storage structures
-
Simplifies scheduling batch jobs because multiple batch jobs that access the
same files can be run concurrently on one or more z/OS images in a Parallel
Sysplex instead of serially on one image
-
Provides the ability to share VSAM data sets at record-level with integrity
and commit and rollback functions for non-CICS applications
-
Offers Backup-while-open to be taken using DFSMSdss and DFSMShsm
-
Enables batch applications to use the same forward recovery logs
-
Helps enable 24x7 CICS Transaction Server (TS) applications
-
Supports System-Managed CF Structure Duplexing
DFSMStvs is an extension to the function provided by VSAM RLS and requires RLS
for accessing VSAM recoverable data sets. VSAM RLS (and therefore, DFSMStvs)
requires the use of a system coupling facility for caching, locking, and buffer
coherency.
ECSA/CSA storage improvements:
DFSMSdfp now supports a new PDSE address space to relieve the ECSA usage.
Error recovery has also been improved to address end-of-task and end-of-memory
terminations for user task abends, operator initiated job cancelations, and
forced address space terminations. These improvements are rolled back to OS/390 V2.10.
DFSMSrmm backup improvements: Users can now run backups of the RMM CDS and
journal at any time, even during inventory management. This enhancement also
enables users to backup and clear just the journal data set to help minimize
the time required. The backup of the journal without the CDS is designed to be
non-intrusive enabling CDS updates to continue throughout the backup.
DFSMSrmm control information: The DFSMSrmm LISTCONTROL command is updated to
display the CDS percentage used and journal status. The operator QUERY ACTIVE
command has been updated with the journal status information. When the status
of the journal changes a new message is issued to notify the operator.
z/OS UNIX System Services
Increase flexibility and reliability in a Parallel Sysplex with Distributed
Byte Range Lock Manager (BRLM):
Moveable BRLM provides the capability of
maintaining the byte range locking history of applications, even when a member
of the sysplex dies. z/OS V1R4 enables distributing the locking history across
all members of the sysplex. As a result, many applications that lock files that
are locally mounted will be unaffected when a remote sysplex member dies.
Movement away from a centralized to a distributed BRLM provides greater
flexibility and reliability.
Shared HFS dead system recovery enhancements: Currently, when a system
goes down, dead system recovery moves file systems that have been defined as
AUTOMOVE=YES to another system in the sysplex. This is done in a random way.
z/OS now provides a method for customers to indicate where file systems should
be placed when a system leaves the sysplex instead of being moved in an
unpredictable fashion. The availability of this function will improve
performance and workload balancing.
In addition, configuration changes can be made to zFS without stopping and
restarting zFS.
Contact z/OS.
Send us your questions and comments.
|
