For additional clarification, consider a batch response time goal of 10 
minutes.  If we assume 5 initiators, and the service class is managed by 
transaction class for these 5 initiators, then how do I not risk have a policy 
adjustment interval where there is only one response time sample in any of the 
buckets during the adjustment interval.  It would render the percentile 
response time worthless.
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to with the message: INFO IBM-MAIN

Reply via email to