Virtual servers; real problems

Chris Procter examines how businesses can realise the benefits of virtualization without risking downtime due to hardware and application failures... by consolidating your servers you have also consolidated any issues that may affect them, increasing the scope of a failure from a single machine to all the virtual machines hosted on the same machine.

Introduction
Probably the biggest of the latest crop of buzzwords is virtualization. The big win of virtualization technologies is the ability to replace many servers with a single box running multiple virtual machines. Most machines have far more resources (CPU, RAM etc.) then they normally need, this is because they are specified by what they are expected to need at their peak use. Careful consolidation of servers onto virtual machines can allow better use to be made of unused capacity which is otherwise only available for SETI research.

Consolidating multiple real servers into a single box has other advantages too, it reduces power consumption and cooling requirements, it simplifies network infrastructure and other such ancillary services (less machines means fewer ports on the switch, fewer KVM ports, fewer racks etc). Management is also simplified, by tying everything together on a single box, resources can be assigned as needed.

However, by moving all your machines onto a single machine you are proverbially putting all you eggs in one basket. Whereas previously a hardware failure would only affect a single machine, now a hardware failure on the physical server will cause a cataclysmic failure as the entire virtual machine dies. So by consolidating your servers you have also increased the scope of a failure from a single machine to all the virtual machines hosted on the same machine.

Read the full article (PDF)

•Date: 4th Sept 2007• Region: UK/World •Type: Article •Topic: IT continuity
Rate this article or make a comment - click here