Originally posted by: mrp
Is there a good way to calculate cache IOPS. something like we do for calculating disk IOPS. How do we how much IOPS can USP with say 20GB cache configured deliver. while we know that a fully configured USP can deliver around 2.5million IOPS, how do we know how much cache IOPS can be possible for our respective USP arrays with 4GB or 10GB or 50GB cache.
also on the same topic i would like to confirm whether the max IOPS of a front-end port on USP is 2500. on what basis is this calculated. does this not differ with 2GB port with the newer 4GB ports ?
it was interesting to look on Graph track for a 9980V that a single production LUN which was published on one of the ports was delivering around 4000 IOPS and i could see that front-end port utilization was around 90%. does this mean that USP front-end port can deliver around 5000 IOPS ?