>10 transactions will be assigned with service class A and 1 transaction will >be assigned with service class B.
>Will this improve the performance of transaction (associated service class >B) by 10 times ? Improve the performance? NO. Based on your definition, each transaction (whether in service class A or B) would be allowed the same number of USING/DELAY samples in proportion to the number of transactions in each service class. For example, consider a velocity goal of 40. Service class A could have 40 USING samples and 60 DELAY samples to meet this goal. Service class B would have 4 USING samples against 6 DELAY samples to be meeting its goals. In each case, WLM would assess goal management based on the number of transactions in each service class. However, the problem is that Service class B might not be able to provide enough samples to represent itself properly. In this case, Service class A might get better performance since it has more units of work contributing samples, and therefore it provides a better view into what is happening within the service class. >Does this mean that A and B will get equal amt of CPU resource which will be >shared by transactions associated with those service classes ? That's not likely. Access to the CPU is determined by dispatching priority and cannot be absolutely determined when two separate service classes are involved. In addition, access to the CPU will be affected by the number of competing service classes at higher importance levels. However, it is reasonably safe to say that Service class B will have more difficulty in being managed since it has only one transaction contributing samples for WLM assessment. Adam ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
