Limited supportBMC provides limited support for this version of the product. As a result, BMC no longer accepts comments in this space. If you encounter problems with the product version or the space, contact BMC Support.BMC recommends upgrading to the latest version of the product. To see documentation for that version, see BMC AMI Ops Monitor for CMF 6.3.

Performance index

Performance index is a relative calculation to determine how well your service classes are meeting their goals.

  • A value of 1.0 shows it is exactly meeting its goal.
  • A value of less than 1.0 shows the service class is exceeding its goal.
  • A value greater than 1.0 shows the service class is not meeting its goal.

The way the performance index is calculated depends on the type of service class.

  • Discretionary Goal

    Service classes whose importance is defined as discretionary do not have goals, so this type of service class does not have a performance index.

  • System Goal

    Service classes whose importance is defined as system do not have goals, so this type of service class does not have a performance index.

  • Execution Velocity Percentage Goal

    For a service class defined with an execution velocity percentage goal, the performance index is calculated by dividing the defined execution velocity percentage goal by the actual execution velocity percentage

    For example, if the goal for a service class is defined as an execution velocity percentage of 90 and the actual velocity percentage is 50, the performance index for that service class is calculated by dividing 90 by 50, for a performance index of 1.8. This would indicate the service class is not meeting its goals, since the performance index is greater than 1.

  • Average Response Time Goal

    The performance index for a service class defined with an average response time goal is calculated by dividing the actual response time by the defined average response time goal.

    For example, if the average actual response time is .25 seconds and the defined average response time goal is .5 seconds, the performance index for that service class is calculated by dividing .25 by .5, for a performance index of .5. This would indicate the service class is exceeding its goals, since the performance index is less than 1.

  • Percentage Response Time Goal

    The performance index for a service class defined with a percentage response time goal is calculated by dividing the actual response time by the response time goal. The actual response time must be calculated by determining which response time bucket (from the SMF 72-3 response time distribution data section) contains the transaction corresponding to the percentage goal. The number of transactions required to meet the goal is calculated as the total number of transactions during the interval multiplied by the percentage portion of the goal.

    For example, suppose a service class has a percentage response time goal of 90% of transactions executing in less than .5 seconds. If the number of transactions for that service class is 50, the goal is for at least 45 transactions (90% of 50) to finish within .5 seconds.

The actual response time is calculated in the following way:

A response time distribution is set up, which contains 14 buckets. In this example, they have the following percentages and values:

  1. count of transactions completed with response times <50% of the goal (less than or equal to .25 seconds): 10
  2. count of transactions completed with response times >50% and less than 60% of the goal (between .25 and .30 seconds): 5
  3. count of transactions completed with response times >60% and less than 70% of the goal (between .30 and .35 seconds): 4
  4. count of transactions completed with response times >70% and less than 80% of the goal (between .35 and .40 seconds): 4
  5. count of transactions completed with response times >80% and less than 90% of the goal (between .40 and .45 seconds): 2
  6. count of transactions completed with response times >90% and less than 100% of the goal (between .45 and .50 seconds): 10
  7. count of transactions completed with response times >100% and less than 110% of the goal (between .50 and .55 seconds): 0
  8. count of transactions completed with response times >110% and less than 120% of the goal (between .55 and .60 seconds): 3
  9. count of transactions completed with response times >120% and less than 130% of the goal (between .60 and .65 seconds): 4
  10. count of transactions completed with response times >130% and less than 140% of the goal (between .65 and .70 seconds): 0
  11. count of transactions completed with response times >140% and less than 150% of the goal (between .70 and .75 seconds): 0
  12. count of transactions completed with response times >150% and less than 200% of the goal (between .75 and 1.0 seconds): 0
  13. count of transactions completed with response times >200% and less than 400% of the goal (between 1.0 and 2.0 seconds): 1
  14. count of transactions completed with response times >400% of the goal (greater than 2.0 seconds): 2

The actual response time is the response time belonging to whichever bucket the 45th fastest transaction fell into. If the 45th fastest transaction finished in .63 seconds, the actual average response time would be .60, because the value of .63 is in the bucket with that value.

In this example, the performance index is calculated as 1.2 (the actual value of .6 divided by the goal value of .5).

Although 80% of the transactions finished in less than .50 seconds (and the average response time for the period may have been well below the goal of .50 seconds), the performance period in this example did not meet its goal, since the goal required that 90% of the transactions complete in less than .50 seconds. The fact that the goal was not met is reflected in the performance index, which is greater than 1.

Related topic

Workload-measurement

 

Tip: For faster searching, add an asterisk to the end of your partial query. Example: cert*