cb-net.co.uk

...WinTel Admin Simplified

  • Increase font size
  • Default font size
  • Decrease font size

Windows NTFS Compression : Decompress Entire Volume

Friday, 02 October 2009 17:51 Chris Bradford
E-mail Print PDF

Windows NTFS Filesystem Compressiopn: Uncompressing An Entire Volume

I recently came across a perofrmance issue on an old x86 WinTel server, The issue, after regular diagnosis showed no obvious cause, appeared to be that the root drive was compressed in order to increase available disk space.

The one problem with NTFS compression is this: 

'When you open a compressed file, Windows automatically decompresses it for you, and when you close the file, Windows compresses it again. This process may decrease your computer performance'(http://support.microsoft.com/kb/307987)

Using the COMPACT command line tool it is possible to both identify all compressed files within a folder and its subfolders. This can be acheived using the command:

compact /I /S

To uncompress all files (assuming you have enough free disk space to do so) you can use the following command to uncompress all compressed files within the current folder and all subdirectories:

compact /U /I /S

After disabling file system compression on the root drive the server is now performing as-expected.

Last Updated on Friday, 02 October 2009 17:59
 

VMWare VCB : Improving Performance of VCB

Sunday, 25 October 2009 21:41 Chris Bradford
E-mail Print PDF

VCB Backup Essentials

Having recently introduced VCB backups into Dataprotector 6.0 I thought I would share a few useful tips for ensuring that backup speeds are as fast as possble.

1) Ensure that all VM's have a scheduled task to zero-out free space prior to VCB running. Windows, when you delete a file does not zero-out the disk space (populate the data blocks with zero's) - soif you had a 20GB drive that contained 15GB of data, then you delete 10GB of data, unoless you zero out this space the backup will still be 15GB.

I use the free 'SDELETE' tool from sysinternals (now Microsoft) to do this, and simply execute a scheduled task before the backup is due to run. SDELETE can eb found here: http://technet.microsoft.com/en-us/sysinternals/bb897443.aspx

2) When running VCB, check the disk queue performance counter on the VCB Proxy server, the storage to which the VCB snapshot is taken can be a serious bottleneck for VCB performance. Initially I was running VCB over fibre, to a fibre attached SAN disk. I found that after 1.97GB the backup would grind to a halt - 200Kb/sec!!! By changing the VCB snapshot drive to local RAID0 storage this increased to over 2.2GB/min, or 37.5MB/sec. Your hardware may be capable of significantly faster speeds.

3) Disable additional disk paths on the VCB Proxy Server: VCB does not like MPIO/multiple paths to LUNS. This step is probably the biggest potential speed gain you'll get. Disable the additional disk objects in Windows device manager, test you backups once complete, if they don't work enable the path you disabled and disable a different one. This can see speed improvements of 100MB/sec.

4) Run multiple VCB snapshots at the same time. Your SAN containing the VM's will, more than likely, support more than 35MB/sec. Just ensure you change the snapshot directory otherwise your backup application may backup multiple snapshots at once!

Last Updated on Monday, 01 March 2010 23:35
 

Resetting iLO Administrator password on HP BL/Proliant Servers

Monday, 01 June 2009 10:21 Chris Bradford
E-mail Print PDF

Resetting iLO Administrator password on HP BL/Proliant Servers

The process below will allow you to reset the ILO/ILO2 Administrator account password from a Windows Operating system running on the server.

Obtain the HP Online ILO configuration tool from here: http://cb-net.co.uk/downloads/HPONCFG.rar

Execute the following command:

 HPONCFG.exe /f Administrator_reset_pw.xml


 

 

Last Updated on Monday, 01 June 2009 10:30
 

VMWare : Troubleshooting VM Performance

Sunday, 25 October 2009 21:54 Chris Bradford
E-mail Print PDF

VMWare Troubleshooting VM Performance in ESX/vSphere

1) Check the number of vCPU's per physical CPU core. VMWare recommened no more than a 5 vCPU : 1 physical CPU core ratio. You can identify contention of CPU resources by running the esxtop command from the shell. Look for constant %RDY above 10% - this indicates CPU contention. First step is to lower the number of active vCPUs configured on the ESX/vSphere server.

For example, if you have 16 cores, the maximum vCPUs that should be defined across all active VMs should not exceed 80.

2) The service console may be low on RAM, you can adjust this by following these instructions: http://www.vmware.com/pdf/esx_performance_tips_tricks.pdf

3) Identify high utilisation VM's. It's better to split functionality out across multiple VM's than it is to configure a single VM with multiple vCPUs.The ESX console will empower you to identify VM's with high CPU/memory urilisation.

4) Identify storage bottlenecks. Look to split your VM's out across multiple LUNs where possible. This will help reduce contention on a particular LUN.

UPDATE 22/02/2010 : Check out the new esxtop article here for further performance troubleshooting tips.

Last Updated on Monday, 22 February 2010 22:27
 


Page 6 of 6

Recent Articles

Most Popular Tags