This post is contributed by Colin Stone.
CICS PA Plugin (Targeted thinking)
In the latest version of the CICS PA plug-in we have attempted to bring a new approach to CICS performance monitoring and assessment. Traditionally users of CICS Performance Analyzer are performance system experts who have honed their skills over many years and have evolved a wealth of knowledge of keeping their CICS systems well balanced and ways of spotting problems before they have a detrimental affect to overall system performance.
We are currently distilling this knowledge and experience into the CICS PA plugin to help new or inexperienced CICS users appreciate more of the factors that influence their CICS systems and what the implications of different aspects of their overall CICS configuration can be.
CICS Statistics – The early warning system
Stats alerts are one of the first points of contact most performance analysts have with their system and generally an alert here requires them to trawl through many different stats records to get a reasonable handle on what is happening inside their system. We have taken some of that experience derived work away from them and now present certain common alerts in a manner that can be more readily consumed or appreciated. CICS PA now presents an interval plot of the users statistics data from before and after an alert. Plotted in this graph are the specific values that contribute the the alert condition they are studying.
Transaction Alert “Maximum tasks reached”
An example is the Transaction Alert “Maximum tasks reached”. This is displayed in the context of: The “max task” value specified, the current and peak active user transactions and the number of times at “max task”. There are additional references and values displayed alongside each interval on the interval plot but the points being shown over a time period give the visual indication of what critically has been happening to the region.
Help me out here..
In conjunction with the interval plot we have put a great deal of effort into the contextual help a user receives. Although the online documentation covers every aspect of their system it is not focussed on the specific situation the analyst is looking with the alert. So rather than the user taking time out to assimilate the online documentation hitting F1 provides the users with the genuine voice of experience that describes the problem domain and what the relevance of these numbers or the rate of change can mean. Not only that it provides users with the practical advise such as understanding areas of caution that need to be considered before going ahead and making changes.
Transaction class “Maximum active transactions in class reached:
The maximum active transactions in class alert is in the same family, as the global transactions alert but in this visualisation we show a breakdown of the transactions that make up that class and the number of transaction attaches. From here the user can link to the equivalent interval plots for both the Applid itself or any specific transaction.
Transaction: “Storage violations for transaction ID”
The final member of this alert family is for individual transaction itself. The Transaction interval plot is shown for a Storage violations for transaction ID alert and these plots can also be selected by menu from the sheet view of the respective stats table. When creating this view the input we received was that it was necessary to look at the transaction attaches in conjunction with the attaches associated with the transaction class if applicable and that of the region overall.
These views have been designed to be navigable and joined up meaning the users are able link from one to another as needed. They show your CICS stats data in a way that has not been shown before letting you make informed decisions about your system supported by real life help and information.