Hey everyone,
I've got a server at the office that was purchased and configured by a former admin, I've been attempting to troubleshoot the problems with the system and have kind of hit a deadend at this point. I'd rather not throw money at it trying to resolve the problems so any advice you can lend would be appreciated.
System Specifications:
Dell PowerEdge 1600SC, updated Bios A10
Dual 2.8ghz Xeon
4gb pc2100 ECC Registered
(3) 36gb Fujitsu 10,000rpm SCSI hdds (RAID 5)
LSI Logic Megaraid 320-1 Controller 64mb PCI-X, 1-internal channel, updated firmware 1L33
Windows 2003 Enterprise Server
The system is solely a Microsoft SQL Server box. That is it's primary and sole function in our datacenter.
I've done some benchmarking on this system as the performance on the drive write speeds just is extremely worrying. I've recently updated the system bios, controller firmware, and formatted and reinstalled Windows 2k3 Enterprise server.
The write speeds are horrible, from the benchmarks I've conducted the read access speeds average around 70mb/sec. However, the write speeds average around 6-7mb/sec. Yes, you heard me correctly when I said 6-7mb/sec. This isn't some phantom number being generated by the benchmark application either. I have thoroughly tested this with real-time extraction benchmarking of rar and zip archives. I have compared the extraction times to a desktop running a single 80gb 7200rpm ATA drive and the desktop outperformed the server by tenfold in extracting and archiving data. I did notice that write-caching is not enabled on this array, could this be the issue with slow performance. I don't think it would be the direct cause, as these speeds are horribly slow and quite warying.
Does anyone know of any issues or array configurations that I may need to check or change to get the performance where it's supposed to be. Could a bad drive cause this type of thing? If so, is there an easy way to check the reliability of the drives? How about memory, would a bad memory module cause something like this? I've tried pulling memory from the system and moving the memory from slot to slot with no luck in resolving the problem.
The scsi connections to the drives and controller seem fine. Is it likely I could have a bad cable? Also, it seems that the former admin purchased a LSI Logic Megaraid 320-1 controller for this system, so the 3 drives are attached to the card directly instead of going to the motherboard. The PowerEdge 1600SC supports SCSI onboard (motherboard) and Dell recommends using the Megaraid 320-0 card which does not have any internal channels, instead it merely uses the card to incorporate the RAID functions into the SCSI interface already built into the system board.
I'm really just looking for any ideas anyone may have in running down the cause / solution of the problems.
I've got a server at the office that was purchased and configured by a former admin, I've been attempting to troubleshoot the problems with the system and have kind of hit a deadend at this point. I'd rather not throw money at it trying to resolve the problems so any advice you can lend would be appreciated.
System Specifications:
Dell PowerEdge 1600SC, updated Bios A10
Dual 2.8ghz Xeon
4gb pc2100 ECC Registered
(3) 36gb Fujitsu 10,000rpm SCSI hdds (RAID 5)
LSI Logic Megaraid 320-1 Controller 64mb PCI-X, 1-internal channel, updated firmware 1L33
Windows 2003 Enterprise Server
The system is solely a Microsoft SQL Server box. That is it's primary and sole function in our datacenter.
I've done some benchmarking on this system as the performance on the drive write speeds just is extremely worrying. I've recently updated the system bios, controller firmware, and formatted and reinstalled Windows 2k3 Enterprise server.
The write speeds are horrible, from the benchmarks I've conducted the read access speeds average around 70mb/sec. However, the write speeds average around 6-7mb/sec. Yes, you heard me correctly when I said 6-7mb/sec. This isn't some phantom number being generated by the benchmark application either. I have thoroughly tested this with real-time extraction benchmarking of rar and zip archives. I have compared the extraction times to a desktop running a single 80gb 7200rpm ATA drive and the desktop outperformed the server by tenfold in extracting and archiving data. I did notice that write-caching is not enabled on this array, could this be the issue with slow performance. I don't think it would be the direct cause, as these speeds are horribly slow and quite warying.
Does anyone know of any issues or array configurations that I may need to check or change to get the performance where it's supposed to be. Could a bad drive cause this type of thing? If so, is there an easy way to check the reliability of the drives? How about memory, would a bad memory module cause something like this? I've tried pulling memory from the system and moving the memory from slot to slot with no luck in resolving the problem.
The scsi connections to the drives and controller seem fine. Is it likely I could have a bad cable? Also, it seems that the former admin purchased a LSI Logic Megaraid 320-1 controller for this system, so the 3 drives are attached to the card directly instead of going to the motherboard. The PowerEdge 1600SC supports SCSI onboard (motherboard) and Dell recommends using the Megaraid 320-0 card which does not have any internal channels, instead it merely uses the card to incorporate the RAID functions into the SCSI interface already built into the system board.
I'm really just looking for any ideas anyone may have in running down the cause / solution of the problems.
Facebook
Twitter