Hello all, There's a lot of back story here, yet I'll try an condense. We're running an OpenE SAN on a Dell 480 <-- think that's correct. OpenE didn't play nicely with the Perc 6i, so we replaced it with an LSI SAS 9265-8I with a GB of NV Cache, which it does seem to like much more and is natively supported in the SAN software. We're running a Hyper-V cluster with SQL 2000 as one of the guests, which is where we're having the issues. I'm running with 4 port MPIO to the SAN device and showing some great performance, yet where we're lacking is SQLIO, which reports high latency on READ functions (85% over 24ms), yet is well within the acceptable range for writes (exceeding the performance of the old physical box). For this I am ruling out the logical disk layout (RAID-50 in two sets striped across 8 disks). Write back is enabled on the card, adaptive read-ahead enabled, disk cache disabled. Latest firmware has been loaded as well. I haven't worked with the LSI cards much, more of an HP person, I'm used to being able to adjust the read/write cache allocation, yet LSI doesn't seem to present this option. Is it there yet hidden somewhere in the CLI? What I'm looking for is anything that I can try to get the read to match the write with this card, any input is appreciated. Thanks all, Keith
Welcome to the forums. Did you try to run the stress tests on hardware using SQLIOSM- http://msdn.microsoft.com/en-us/library/bb649502(v=sql.90).aspx fyi. For the hardware I always refer to the vendor's documentation or contact them to if they can confirm whether the task can be accomplished or not.