Server Performance Analysis

Managing response time and meeting capacity needs, without costly "over-engineering" is called performance management.  Performance management is a way to figure out the most cost-effective way to provide for the service requirements of the organization.  Plans for growth in service requirements in terms of additional users, new applications or efforts to improve existing service levels all effect server performance.  Performance monitoring ensures proper system operation and out-of-control software.

Building blocks for server performance include:  

  - continuous measurement process

  - an automated measurement infrastructure

   -application-oriented response time measurements

View Daily Graphs

With a continuous measurement in process is in place, data is collected 24 hours a day, seven days a week, not just when problems arise.  To be useful, the data must be summarized, analyzed and reported on.

 Performance management encompasses:

 - solving day-to-day problems

-  tracking long term performance

-  predicting the impact of change

-  communicating to management and users

Server performance analysis is different than real-time monitoring.  Real-time monitoring  is used to monitor the server for pre-defined conditions - i.e. a file system filling up, CPU utilization above a defined threshold, disks having large service times, excessive paging.  

Patrol Perform/Predict is the tool used by IS for server performance.  Perform/Predict is capable of collecting and summarizing performance data and creating graphs for long range planning and analysis.   By monitoring a server while application testing is performed, Perform/Predict can be used to forecast the impact of the application when it is implemented into production.