Tivoli Performance Analyzer - for Forecasting and Predictive Analysis
Kiran Shashi 2700069WSG Comments (3) Visits (3924)
The IBM Tivoli Performance Analyzer also called as ITPA - is an Tivoli agent which looks at the Tivoli Data Warehouse, and analyzes the historical data from various agents and comes up with a trend and also forecasts how certain resources are going to look like in future.
Let me just break up the above a bit.. The Performance Agent is another Tivoli agent, and is capable of analyzing historical data from several agents stored in Tivoli Data Warehouse ( The agents could be it OS, ITCAM, VMware Agent, Oracle and the DB2 agent.)
Currently the ITPA supports the above six agents.
The six agents have to be configured for historical data collections ( be it hourly. daily, weekly, quarterly and yearly ) in TDW, for the performance Analyzer to view and come up with the trend and forecast.
The ITPA has pre-defined tasks for each agent ( or domain ) that gets installed as part of the Out-Of-the-box domain installation, and are configured to begin running once ITPA is started.
The user however, has the ability to create new tasks , or alter the existing tasks on how to and when to calculate the data.
ITPA renders this trend and forecast on the Tivoli Enterprise Portal GUI or using TCR reports. User can navigate through the different tasks and view how the agent's resources ( for e.g: Disk , CPU in case of OS agent, OR Web Response Time or Robotic Response Time in case of ITCAM ) have performed, and how it would look like in future.if the current trend continues - thereby giving the system admin or engineers managing the infrastructure some indication as how to how to plan for additional hardware or infrastructure deployment.
ITPA supports analytical modules called Linear Trending and Non-Linear Trending data calculation, as well as create analytical module based on expressions on the agent's attribute groups, that would best fit their business needs as well.
Linear trending is based on Linear Regression algorithm and looks at how the data has been over a course of time and comes up with a 'best fit' line or trend.
The Non-Linear calculation is through another plugin called SPSS, which is installed on top of ITPA, and comes up with an even better prediction as its algorithms are more optimized.
So, while the Linear Trending is incorporated into ITPA by default, the Non-Linear trending support comes in by installing the SPSS add-on-. The user can look at both to come up with best guess answer and plan his resources.