Brian, when you say 'relative performance metrics' are you asking about the CPU cost associated with the various paths? If so, I don't think I have any current data on that. Partly because, it does depend on a lot of different things. My suggestion there is to find a virtual machine that is fairly 'typical' and using the methods Barton & Rob described get the r/w view. Then do an experiment with and without MDC. The tricky part here, as was pointed out, is shared minidisks. Look at the CP CPU time for the guest with and without, prorated on a per virtual I/O.
I should mention that they did pay a lot of attention to pathlengths when MDC was implemented, so I would not expect the difference to be significant in terms of processor time. At least not enough to make the time it takes to continue to do that type of analysis worthwhile. I'm sure there are extreme cases (such as using mapped minidisks to dataspaces and MDC at the same time) where costs are noticeable. So, what we have found more effective is go after the big hitters, such as read-once/write-once, write-only disks, etc. and make sure MDC is off for those. Rob, this is the first I've heard that the insert (not replace) was happening in general on write I/Os. I knew there were exceptions, but this seems broader. I'll put it on my list to investigate. Do you recall if that was using diagnose I/O or SSCH? If diagnose, was it recordlevel or default? Thanks. Bill Bitner - VM Performance Evaluation - IBM Endicott - 607-429-3286
