No longer is virtualisation restricted to the "low hanging fruit" - non-critical, low end servers that can easily be consolidated and at low risk. People are now looking to embark on the next stage and virtualise their business critical applications. Increasingly, a virtual machine will be the default home for a new application.
As it becomes more pervasive, virtualisation changes the dynamics of the data centre. In many large organisations, it takes many weeks or even months to get a new server installed. Even when the hardware procurement process is pretty streamlined, there are still delays in waiting for the kit to be delivered, finding a suitable change window when it can be installed, finding space for it, connecting it to the power supply, getting it hooked up to the network(s) and perhaps to storage.
By contrast, a virtual server can be created almost instantly.
Not surprisingly, this can stress some operational processes which have previously relied on the server delivery bottleneck, together with space and power constraints to provide a natural buffer to otherwise unbounded demands for more servers.
What happens in the longer term, once these issues have been ironed out? I'm assuming that in many cases, the number of physical servers in the estate will be reduced. That's part of the justification for the virtualisation exercise in the first place. But what about the total number of servers (virtual and physical)?
I would be interested to see statistics on the number of virtual servers that enterprises end up with, some years down the line. Does this rapidly exceed the original number of physical servers? And how do the long-term growth rates compare to those for the physical estate prior to the virtualisation exercise?
Does anyone have any data they are willing to share?
Posts
No comments:
Post a Comment