HiperLogic

Virtualization, High Performance Computing, Healthcare IT, Enterprise Computing

June 2010

You are currently browsing the monthly archive for June 2010.

While vSphere 4.1 is not out yet, there are numerous leaks and announcements on the forums and other blogs on what to expect in 4.1

A partial list of features expected include:

1. ESX and ESXi direct support for Microsoft Active Directory through Likewise technology
2. Memory Compression for even higher VM density per host.
3. Scripted installations for ESXi making large rapid deployments supported and quick.
4. Storage I/O Control – Think of this is shares and limits for storage I/O to make sure important VMs get priority.
5. DRS VM Affinity – Restrict VM placement to particular hosts inside of an ESX cluster.
6. New HA dashboard for monitoring your ESX HA cluster.
7. Up to 8 simultaneous VMotions.
8. Microsoft Volume Shadow Service (VSS) in Windows Server 2008 and 2008 R2 guest operating systems for VADP
9. QoS engine that can prioritize between virtual machines, vMotion, Fault Tolerance (FT) and IP storage traffic.
10. Support for USB pass-through

Should be a very nice release!

The new Veeam Sure Backup Technology that checks that your VM will be able to recover from a backup is shown in this video. VERY cool, good example of how you can backup a system that is corrupted and have a false sense of security that it can be recovered without the use of SureBackup:

Storage I/O Control is a new feature coming in vSphere 4.1 that will allow the admin to control and throttle VM throughput to the datastores when a VMs latency crosses a certain threshold.

This is an entertaining video made by someone else that demonstrates this ( pre-release ) technology:

This is a great site for keeping track of VMware build information:
VMware Release Build Information

Tags:

[vSphere 4.0 Update 2 is out which includes updates for ESX/ESXi hosts (build 261974) as well as the vCenter Server (build 258672).

This update primarily has more hardware support and guest OS support, and the ability to see more performance statistics for NFS datastores, along with the usual bug fixes.

To see the new NFS stats use (r)esxtop and hit "u" to see the NFS datastore stats like below:

esxtop

esxtop

NOTE: If you are using VMware View, do NOT upgrade at this time. See this article for the latest info on this issue: 1022830

UPDATE June 14th: If you are using EMC CLARiiON, make sure and read This EMC Post Before upgrading.

Update July 9th: Another nice feature in this update is how “Delete All” snapshots is handled, reducing the amount of space required and lowering the chance you fill up the datastore. See Yellow Bricks for more details

HiperLogic offers hardware solutions from HP ( and other vendors ), and for Virtualization we are a big fan of the HP DL380 G6/G7. We have noted during our VMware health check service that many DL380 installs that were done by our competitors were not ordered and/or configured for maximum performance. This post is to assist customers in getting it right.

Memory
Not having an optimal memory configuration can drastically reduce performance. HP has engineered the DL380 G6 to support large amounts of RAM and still keep the memory speed at 1333MHz (With 95W CPUs ). The memory configuration on new Nehalem based systems is no longer a simple matter of just picking a capacity and running with it like in the old days. If you want to configure your own memory see www.hp.com/go/ddr3memory-configurator

Disk

Next is disk performance. You should purchase a Battery Backed Write Cache Module with any DL380 G6 for vSphere ESXi or you will have performance problems, see this HP advisory. Even if you are running ESX today, ESXi may in the very near future be the only path forward, so better to spend a few dollars to future proof your investment.

BIOS

Having the wrong BIOS settings can really make a huge difference in performance. For example you may consider setting:

Turbo Boost On
HP Power Profile to Maximum Performance
HP Power Regulator to Static High Performance
Thermal Configuration Increased Cooling

Local Disk

HiperLogic also likes to have some fast local disk, the DL380G6 has plenty of internal storage for putting lots of fast disk. Having fast local disk is great even if you use a SAN as you can always storage vmotion to a local VMFS volume when testing or debugging VM disk I/O or VM disk performance issues, and for temporarily working around SAN issues etc. Disk is cheap, and it is nice to have the flexibility.

Finally always check the VMware HCL, for the DL380 G6 that is here

There is also a HP VMware HCL and vSphere 5 for HP Servers

Tags:

Rodney Mach of HiperLogic was selected as one of the VMware vExperts for 2010.

The VMware vExpert Award is given to individuals who have significantly contributed to the community of VMware users over the past year. vExperts are book authors, bloggers, VMUG leaders, tool builders, and other IT professionals who share their knowledge and passion with others. These vExperts have gone above and beyond their day jobs to share their technical expertise and communicate the value of VMware and virtualization to their colleagues and community.

Rodney runs the Ann Arbor Virtualization User Group ( http://aavug.org) and blogs here on the HiperLogic blog http://hiperlogic.com/blog and at http://rodmach.com/blog on Virtualization technologies. Thanks to VMware for the honor!

Recently a customer noted that many of their VMs were still showing up on an old datastore after doing a storage vMotion to a new datastore. Browsing the actual datastore showed the VM was not there, but the VIC continued to report it was.

The issue was the customer had Storage vMotioned with ISO images connected, which you should not do. The simple fix was to disconnect all the ISO images from those VMs. This is easy through the VIC GUI, but to automate it and make sure none were missed we used Power CLI:

Get-VM  | ForEach ( $_ ) { Get-CDDrive $_ | Where { $_.IsoPath.Length -gt 0 -OR $_.HostDevice.Length -gt 0 } | Set-CDDrive -NoMedia -Confirm:$False }

Or to just do this for a particular datastore:

Get-VM  -DataStore mydatastore | ForEach ( $_ ) { Get-CDDrive $_ | Where { $_.IsoPath.Length -gt 0 -OR $_.HostDevice.Length -gt 0 } | Set-CDDrive -NoMedia -Confirm:$False }

To restart HPC Server 2008 specific processes on compute nodes using just Powershell:

$nodes = Get-HpcNode
$services = "HpcManagement", "HpcNodeManager", "msmpi"

foreach ($node in $nodes) {
  foreach ($service in $services) {
     (gwmi win32_service -filter "Name='$service'" -computername "$node").stopservice()
     (gwmi win32_service -filter "Name='$service'" -computername "$node").startservice()
  }
}

© 2006-2010 HiperLogic, LLC.  |  Serving the Ann Arbor, Southeast Michigan, and Ohio region  |  (888)-268-3930  |