>Moving on, it has been demonstrated that the CF engine needs to be reasonably >close to the coupled engine speed else there is a risk of capacity loss.
Define 'reasonably close'. Define 'a risk'. We had 9672-C04 CFs with 9672-RXn & z/900 images. Yes, this was a few years ago. Due to budgetting constraints, it was going to be a year before the CFs were to be upgraded. An IBM'r (since retired) came in and helped us with the analysis. It came out that we would save 100ms on an IMS transaction, if we were to upgrade. Since they were synchronous locks, that was the equivalent of 100ms of CPU time, differing percentages of the processor, based on whether it was a 9672 or a z/900. This worked out to a cost of 2% of the smallest processor. We presented the results to management. >It is also true that lots of other things can affect this, such as the >oft-mentioned Sync to Async conversion algorithms, distance, Processor and s/w upgrade costs. Our management decided that the 2% hit was 'cheaper' than upgrading. We waited the year. As a performance analyst, you want the fastest and the bestest. As a capacity analyst, you have to take business need, budgetting, and other financial issues into account. "It's been demnonstrated ..." does not sell business cases very often, if at all. - Too busy driving to stop for gas! ---------------------------------------------------------------------- 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
