Proliant ML330 Hard Drive Performance Boost
At work, we're using an old version of Webtrends to do all our web analytics. With average daily log files of 2 GB, this is a pretty intensive procedure that taxes all aspects of the system like memory, disk, and CPU. I have been running these stats on a base desktop system with 2.8 GHz Pentium 4 and 512 MB RAM.
Certainly not the best machine for Webtrends but it has been adequate for the past year. However, I recently was forced to replace this setup with a Proliant ML330 that was a decommissioned webserver. With a 2.8 GHz Xeon and ATA-RAID drives, I expected this to be a good base once the 256 MB of memory was increased.
What I didn't expect was for my 2 day-long reports to take double that time on the Proliant. Even with the memory starved machine, this was totally unexpected. Further investigation revealed the disk utilization was where the sluggishness was most apparent.
After performance testing within Windows and off a boot CD, I determined the striped hard drives (RAID-0) were performing at a quarter the speed of the desktop I tested. After opening the chassis, I found the drives were both connected to one channel on the built-in RAID.
I decided to put the second drive on its own channel, which is my preferred setup anyway, and retested the system. It is now performing approximately 50% faster than the desktop benchmark.
Certainly not the best machine for Webtrends but it has been adequate for the past year. However, I recently was forced to replace this setup with a Proliant ML330 that was a decommissioned webserver. With a 2.8 GHz Xeon and ATA-RAID drives, I expected this to be a good base once the 256 MB of memory was increased.
What I didn't expect was for my 2 day-long reports to take double that time on the Proliant. Even with the memory starved machine, this was totally unexpected. Further investigation revealed the disk utilization was where the sluggishness was most apparent.
After performance testing within Windows and off a boot CD, I determined the striped hard drives (RAID-0) were performing at a quarter the speed of the desktop I tested. After opening the chassis, I found the drives were both connected to one channel on the built-in RAID.
I decided to put the second drive on its own channel, which is my preferred setup anyway, and retested the system. It is now performing approximately 50% faster than the desktop benchmark.