Virtualisation is clearly one of IT’s great fashions, and as such continues to attract huge amounts of interest. Unlike many technology-based fashions – and we can all remember quite a few – virtualisation solutions are being deployed in anger in many areas, especially in the world of x86 servers. As is well appreciated, getting budget approval for any project can be problematic – so just how do the economics of “virtualisation” stack up?
We know from research that in the majority of projects cost considerations are always to be found at the centre of things. The area in which “virtualisation” has achieved its greatest degree of deployment supporting live solutions is in x86 server consolidation projects. In many instances the cost arguments have been relatively straightforward and usually centre around easy to calculate arguments involving raising server utilisation rates, lowering the consumption of electricity and reductions in software licence costs through running applications on smaller numbers of servers.
Much of the obvious low-hanging fruit in x86 server virtualisation has now been picked, and we are beginning to see attention focus on the operational cost impacts of running virtualised systems. Once again,Reg readers have told us they are now beginning to experience challenges with the daily management of virtualised server environments.
Typical challenges mentioned include virtual machine sprawl whereby large numbers of new virtual machines can be deployed very easily.
“Where we previously had server sprawl we now have virtualised instance OS sprawl and it happens at 100 times the rate,” one reader told us.
Without good processes in place to control the generation of such VMs it can be alarmingly easy to run into problems with software licensing, or simply running out of physical resources.
The ease of creation appears to encourage virtual machine deployment making it imperative that some form of braking mechanism be put in place to encourage users of virtual servers to take them down when no longer actively used. This is an area where either new forms of change management processes must be put in place or the use of some variety of charge-back accounting be used to throttle resource consumption.
Such aspects will inevitably have a cost impact: “Your costs will not go down,” says another reader. “Most costs were in the management and people anyway and that has not changed.”
Beyond x86 server virtualisation there is growing evidence that attempts to create business cases for undertaken virtualisation projects in areas such as desktop and storage frequently run into trouble. Indeed, in a survey during the summer nearly 40 per cent of Reg readers answering questions on desktop virtualisation highlighted the lack of a strong business case as a major or significant blocker to such projects. This coupled with the usual questions of acquisition and infrastructure impact costs dominated the list of inhibitors, clearly blocking many potential projects from getting started.
A similar situation exists in the area of storage virtualisation. It raises the question of whether the difficulties in creating viable business case for such projects is linked with the lower level of understanding of the associated technologies plays a part in working out in which areas such projects make sense and what solutions best fit specific business needs.