CCSS Adds Predictive Capability to System Monitoring Tool

August 7, 2007 Alex Woodie

Organizations that purchase CCSS‘ QSystem Monitor version 12 to monitor the real-time performance of their iSeries and System i servers received a little bonus last month. That’s when the i5/OS software specialist announced the inclusion of a new feature to V12 that shows what the future holds across a variety of i-server performance metrics, from disk and CPU usage to transactions per minute and new orders.

QSystem Monitor is a client/server application that shows system administrators how their System i and iSeries servers are using system resources. Server-side agents monitor specific resources, while the results are displayed on the graphical Windows client, called the Online Monitor.

It’s been almost a year since CCSS announced QSystem Monitor version 12, by all accounts a landmark release of the product. Previous versions of QSystem Monitor contained a “rigid framework” that could only monitor certain performance metrics, according to product manager Paul Ratchford. But with 100 new agents, the new version dramatically opened up the system monitoring possibilities, while also giving users more control over how data is presented in the Online Monitor.

The overhaul of QSystem Monitor was still playing out five months ago, when CCSS added new DASD monitoring capabilities. The new feature gave users more granular coverage of their DASD usage, including the capability to slice and dice a number of aspects of Auxiliary Storage Pool (ASP) usage. CCSS didn’t label it a point release, just a new feature for V12 users–an increasingly common approach to releasing new features among ISVs.

Last month CCSS did it again when it when it announced the new predictive capability as a new feature available to V12 users. The new feature, which CCSS says was not available when V12 first shipped nearly a year ago, enables users to see how various performance metrics will change over time. The product bases its predictions on the historical usage of various system resources and metrics–from DASD usage to job growth, from transactions per minute (TPM) to the number of new orders entered into the system.

The predictive tool shares its findings via a graphical report in the Online Monitor that is easily shared during budget meetings, Ratchford says. “It’s a very useful tool to be able to say, based on our past-use history, we will effectively run out of disk on this day in the future unless we implement a clean up,” he says. Managers can choose between two views for disk usage, one that show daily use in a monthly period, and another that shows monthly use in an annual period.

While most users are likely to use the new predictive capability to forecast future DASD usage, the new feature can be applied to “virtually all system elements,” a CCSS spokesperson says. “It also has quite different applications, from general business metrics to future planning for resources,” the spokesperson says. “We’ve made a separate release of it because it’s a unique feature and typically not something people would expect from a system monitoring solution.”

For example, managers could use the tool to generate a preview of quarterly or year-end sales figures by counting the number of new orders entered into the system. The tool can be configured to count only files of a certain type from specific business applications, giving managers another reporting method. Given QSystem Monitor’s history as a data center tool, however, it’s unlikely to supplant Query/400 anytime soon.

Aside from predicting DASD doomsday scenarios (running out of disk on this machine is very, very bad, but you probably already knew that), the new predictive capability in QSystem Monitor can be used to keep a watchful eye on a variety of other System i trouble spots, including jobs that consume more than their share of CPW.

CCSS says one such trouble spot is QZDASOINIT jobs, which are basically placeholder jobs that i5/OS starts and keeps running to cut down on job initialization time when new data transfer requests are made, usually from PCs attached to the iSeries or System i server. However, because QZDASOINIT jobs often consume large amounts of CPW (and don’t, in and of themselves, do anything), they can reduce processing power available to higher priority jobs. Therefore, predicting the future growth of QZDASOINIT jobs can help administrators balance response times across all the applications they’re responsible for.

CCSS also lets administrators implement real-time thresholds for specific resources. If these thresholds are exceeded, they automatically generate an alert that’s sent to the administrator. CCSS says these alerts are handy for spotting dramatic unforeseen situations that interrupt the projection of normal usage, and which therefore require immediate attention to forestall a catastrophe.