Disk array performance

[steve_26]

I have just installed two identical rp4440 servers connected via SAN to an EMC CX300 Array. I am benchmarking performance and am surprised by what I've seen so far. One server will be for running a Production Sybase database server the other for Test. The CX300 contains 45 73GB 10000RPM disks. I have allocated 32 of the disks to Production in 4 8-disk raid 10 groups. I have attempted to spread the Sybase data evenly across the raid groups. The Test server database is allocated on a 6+1 Raid5 group. (Total databases about 133GB currently.) I carved off 9 luns for production vs. 2 for Test. LV allocations for the two servers are identical with 9 each (same size, same database allocations,all raw). What surprises me is the Test server is outperforming Production. I am running a nightly warehouse build process which runs about 7 hours on our current system and is around 3 hours on the new ones. There is a lot of write activity involved including database logging. I've just started investigating and know you don't have all the details, but at face value can anyone explain the small raid 5 group outperforming the raid 10? (2 hours 20 min. vs 2 hours 50 min.)

[Alzhy]

1. You probably have a better tuned DB instance on your test environment that's using RAID5.

2. Your datafiles on your test environment are probably better laid out.