Perfmon to analyse Disk Behaviour on window server 2008
Posted on 2013-06-25
We have one central SCCM Site server where we are facing an issue. The SMS_Inventory_Data_Loader of central SCCM Server unable to move the MIF Files to the database , It resulted in piling up the dataldr.box folder with huge *.MIF files.
I am not an SCCM Administrator , but when asked the administrator he said *.mif are inventory files which moved to database from dataldr.box location with the help of SMS_Inventory_Data_Loader .
As the data is not getting moved to the database , most of the inventories are not getting updated.
Hence we have opened the call with Microsoft regarding this issue and they have come up with an answer stating there are some disk issues in a server
Basically we are using one physical disk (G:) for SCCM and (F:) for SQL ,
The Technician analyzed the perfmon logs of physical disk and found the Disks needs to updated with latest drives and defragmentation has to performed.
Below is the analyses.
\PhysicalDisk(2 F:)\% Idle Time
[Min: 0%, Max: 102.73%, Avg: 52.84%]
- Disk is heavily utilized and may be a bottleneck.
- Investigate Disk bytes/sec and individual process IO Data bytes/sec for high disk usage.
\PhysicalDisk(2 F:)\Avg. Disk Queue Length
[Min: 0, Max: 4,766.439, Avg: 17.676]
- Generally up to 2 per spindle (Hard Disk Device) is acceptable.
\PhysicalDisk(3 G:)\Split IO/Sec
[Min: 0, Max: 5,114.549, Avg: 6.405]
- Split IO operations make up 9.01% of Disk Transfers.
- Avg should be extremely close to 0.000.
- RAID may be too small/NTFS block too small.
- Disk may have moderate to heavy fragmentation
Action Plan for disks
1. Could you please involve vendor for the disk to check it? May be updating the drivers may help.
2. There are lot of Split I/O for both F: disk and G: disk. I would recommend to defragment the disks
Also he was talking about Read Cache and write Cache , can some one tell me what are these all about.
Please let me know the above statement made by technician are vaild , since I have never taken or analysed perfmon logs.
Also , Can someone let me know how to take the perfmon log and analyse it?
I want to be 100% before I carry out the "Action Plan for Disk" provided by the technician