alex.angel
n00b
- Joined
- Nov 18, 2004
- Messages
- 10
Build following system at work (I'm a junior SA on co-op placement):
Dell Precision 410
-Pentium 2, 256MB RAM, 10/100 onboard NIC
-4x Maxtor Maxline 250GB SATA 7200RPM drives
-Promise 2410SA sata-raid controller
Built the array as RAID5, 64kb cluster size, using all 4 disks. Made a 20GB partition for Windows 2003 Enterprise Server, with the remaining (bit more than 650GB) as an empty ntfs partition.
Wanted to check i/o perf to make sure the array was up to snuff, and I get...crap. 26MB/s write in SiSoft Sandra drive benchmark (latest) and 65MB/s read. These results are awful! I might've chalked them up to "software RAID on a slow cpu" but the 2410SA is a hardware RAID card so this doesn't fly. I checked to make sure it's not a compressed filesystem, enabling/disabling write caching doesn't affect the results. I tried the latest winbench (as storagereview.com uses for their reviews) and the server benchmark pegs me lower than a single 80GB drive!
My boss wants to know what the hold-up is with my storage array work .this machine will let me test out some fancy-shmancy network storage system (ok nothing fancy, just DFS) and once I work the kinks out we're going to a much, much larger storage setup (thinking of using a few Promise 15-drive SATA-SCSI 3U boxen on a SCSI card) but I'm stymied by this performance anomaly. Any help/suggestions?
Dell Precision 410
-Pentium 2, 256MB RAM, 10/100 onboard NIC
-4x Maxtor Maxline 250GB SATA 7200RPM drives
-Promise 2410SA sata-raid controller
Built the array as RAID5, 64kb cluster size, using all 4 disks. Made a 20GB partition for Windows 2003 Enterprise Server, with the remaining (bit more than 650GB) as an empty ntfs partition.
Wanted to check i/o perf to make sure the array was up to snuff, and I get...crap. 26MB/s write in SiSoft Sandra drive benchmark (latest) and 65MB/s read. These results are awful! I might've chalked them up to "software RAID on a slow cpu" but the 2410SA is a hardware RAID card so this doesn't fly. I checked to make sure it's not a compressed filesystem, enabling/disabling write caching doesn't affect the results. I tried the latest winbench (as storagereview.com uses for their reviews) and the server benchmark pegs me lower than a single 80GB drive!
My boss wants to know what the hold-up is with my storage array work .this machine will let me test out some fancy-shmancy network storage system (ok nothing fancy, just DFS) and once I work the kinks out we're going to a much, much larger storage setup (thinking of using a few Promise 15-drive SATA-SCSI 3U boxen on a SCSI card) but I'm stymied by this performance anomaly. Any help/suggestions?