IBM ZOS V1.12 Manual - page 19
IBM United States Software Announcement
210-008
IBM is a registered trademark of International Business Machines Corporation
19
the system to analyze a component or a subsystem that it suspects is failing and
provide warnings and guidance for operators and system programmers.
• z/OS extends its high-availability characteristics by going beyond failure detection
to predicting problems before they occur. With Predictive Failure Analysis (PFA),
introduced with z/OS V1.11, your z/OS system is designed to learn heuristically
from its own environment to anticipate and report on system abnormalities. It
compares present and past behaviors and models system behavior in the future,
and is intended to notify you when a system trend might cause a problem.
For z/OS V1.12 PFA is planned to monitor the rate at which the system is
generating SMF records. When the rate is abnormally high for a particular system,
the system will be designed to issue an alert warning you of a possible problem,
potentially avoiding an outage. PFA can take into account the normal swings of
daily, weekly, or monthly spikes and can learn the idiosyncrasies of your system,
thus avoiding false warnings given by static monitors.
• In z/OS V1.12, a new capability, z/OS Run Time Diagnostics, is planned to help
when the need for quick decision-making is required. With Run Time Diagnostics,
your z/OS system will be designed to analyze key system indicators of a running
system. The goal is to help you identify the root of problems that cause system
degradation on systems that are still responsive to operator commands. Run Time
Diagnostics is anticipated to run quickly to return results fast enough to aid you in
making decisions about alternative corrective actions and facilitate high levels of
system and application availability.
• In z/OS V1.12, a new Timed Auto Reply Function is planned to enable the system
to respond automatically to write to operator with reply (WTOR) messages. This
new function is expected to help provide a timely response to WTORs and help
prevent delayed responses from causing system problems.
• z/OS availability is beyond the server as well. Parallel Sysplex can provide a large
single system image, dynamic load balancing, fault tolerance, and automatic
restart capabilities, so a single cluster can be used for scalability and performance
as well as for availability and disaster recovery. With z/OS V1.12, Parallel
Sysplex technology is planned to be updated with new health checks; improved
command routing; and improved network traffic routing, security, availability and
reporting. There are also plans to provide autonomics whereby the z/OS system
can help identify CF structures and network connections that are unresponsive
or in a degraded state. In addition, GRS and XCF components are planned to
automatically initiate actions to preserve sysplex availability to help reduce the
incidence of sysplex-wide problems that can result from unresponsive critical
components.
Details on availability improvements planned for z/OS V1.12 include:
• Over time, VSAM key-sequenced data sets (KSDSs) for which records are added
and deleted have often become fragmented and have a significant number of
empty Control Areas (CAs) that consume DASD space, increase the size of the
indexes, and reduce performance. Performance and DASD space utilization can
usually be improved for such data sets by copying, deleting and reallocating,
and reloading them. This requires scheduled outages for applications using these
data sets. In z/OS V1.12, DFSMSdfp is planned to allow you to specify that VSAM
dynamically reclaim unused control areas for KSDSs, including those used for
record-level sharing (RLS), and reclaim the associated index records as needed.
This new function is intended to help you preserve performance, minimize space
utilization for KSDSs, and improve application availability, and to allow you to
discontinue the use of jobs whose sole purpose is to reorganize KSDSs.
• In z/OS V1.12, a new component named z/OS Run Time Diagnostics is planned.
This function is planned to help you reduce the time spent deciding what actions
to take to resolve a problem. It can be used to identify potentially related
symptoms and causes when it appears a significant system problem might affect
the systems ability to process your workloads. Often, you must quickly analyze
these problems to preserve application availability. Run Time Diagnostics is
designed to run using the START operator command and return results quickly to
help you decide alternative corrective actions and maintain high levels of system
and application availability. Run Time Diagnostics is planned to identify critical
messages, search for serialization contention, find address spaces consuming