B.3.1 Training Period - IBM z13s Technical Manual

IBM zAware detects conditions that typical monitoring systems miss because of these
challenges:
Message suppression (message too common): Common messages are useful for
long-term health issues.
Uniqueness (message not common enough): These messages are useful for real-time
event diagnostic procedures.
IBM zAware assigns a color to an interval based on the distribution of interval score:
Blue (Normal)
Interval score between 1- 99.5
Gold (Interesting)
Interval score between 99.6 - 100
Orange (Rare)
An interval score of 101

B.3.1 Training period

The IBM zAware server starts receiving current data from the z/OS system logger that runs
on z/OS monitored clients and from Linux SYSLOG for Linux on z Systems monitored clients.
However, the server cannot use this data for analysis until a model of normal system behavior
exists.
The minimum amount of data for building the most accurate models is 90 days of data for
each client. By default, training automatically runs every 30 days. You can modify the number
of days that are required for this training period, based on your knowledge of the workloads
that run on z/OS monitored clients. This training period applies for all monitored clients.
Different training periods cannot be defined for each client.
B.3.2 Priming IBM zAware
Instead of waiting for the IBM zAware server to collect data over the course of the training
period, you can
and requesting that the server build a model for each client from the transferred data.
Currently, the bulk transfer of Linux historical data is not supported.
B.3.3 IBM zAware ignore message support
When a new workload is added to a system that is monitored by IBM zAware or is moved to a
different system, it often generates messages that are not part of that system's model.
Subsequently, these messages are flagged as anomalous and cause orange bars to appear
on the IBM zAware analysis window.
Sometimes, the reporting of anomalous behavior is caused solely by the new workload, but
sometimes a real problem is present as well. Therefore, it is not appropriate to automatically
mark all the messages as "normal" when new workloads are introduced. IBM zAware
provides the ignore message support to give you input into the IBM zAware rules. This
function allows you to mark messages as "ignore." An ignored message is not part of the IBM
zAware interval anomaly scoring, but it does appear in the output.
prime
the server. You do so by transferring prior data for monitored clients
Appendix B. IBM z Systems Advanced Workload Analysis Reporter (IBM zAware)
463