Is dynamic power management impacting the hardware MTBF?

Posted by Alessandro Perilli   |   Monday, December 01, 2008   |   2 Comments

There’s no doubt that hardware virtualization has a chance to significantly reduce the power and cooling usage in most companies thanks to remarkable server consolidation ratios. But some vendors are trying to use automation to make the IT greener (as marketing people love to say).

Virtual Iron and VMware now offer the capability to consolidate into a single server the virtual machines served across a bunch of virtualization hosts.
The VMs are live migrated and the empty virtualization hosts are powered off until there is an actual need for them.

While this sounds a great thing it may have some side effects that few companies are considering.

Chris Wolf, Senior Analyst at Burton Group, is wondering if this dynamic power management has a concrete impact on the hardware mean time between failure (MTBF).

Interesting enough the major IHVs that he contacted didn’t perform any test to find out.

Labels: ,

2 Comments

Anonymous anykey Tuesday, December 02, 2008 12:14:00 AM  
usually only harddisks suffer from being turned on and off, and because of that reduce their life expectancy by 1/3, but as with most virtualization implementations, servers have no local storage for data and over more often no hard disks at all. So my guess would be that this would not have any impact on MTBF.
Blogger nosatalian Thursday, December 04, 2008 1:53:00 AM  
What he said <--

What exactly is going to fail by turning on and off? Perhaps some fans might flake out earlier, but I don't think it is a major concern. Anyways, the point of virtualization is that an individual hardware failure/downtime doesn't affect the workload. I get the feeling most of these "concerns" over features such as VMotion and DPM are FUD put out to buy time for Microsoft to catch up in feature parity.

Add New Comment