Life beyond server consolidation
Hurdles to the virtual world
Workshop Virtualisation software vendors have in their marketing kit bags a very simple chart, which comprises three circles.
The first, on the bottom left shows ‘simple’ server virtualisation, that is, use of virtual machines to consolidate multiple workloads onto a smaller set of servers.
In the centre is ‘managed’ virtualisation, in which IT operations can provision new virtual machines as and when needed. And on the top right we have some state of fully dynamic, flexible, agile, cloud-based, choose-your-utopia IT that works with minimal human intervention. I forget the exact terminology but I hope you get the picture.
This vision of virtualisation as a journey is compelling, not just for virtualisation software vendors but also for IT managers. We hear of ops staff concerns about being made redundant by this dynamic nirvana, which has an element of truth (particularly if shift pay is involved). The bigger truth however, is that in most IT departments, people have better things to be getting on with than responding to the latest unexpected deployment request. Even where space and power are not at a premium, it can be a challenge to fit in new equipment for example – so, what’s not to like about virtualising it all? Are most organisations destined to head towards higher orders of virtualisation beyond simple consolidation?
According to recent research we have undertaken, the challenge comes not with virtualisation per se, but with its management. Consolidation-by-virtualisation requires little change operationally. Indeed, for those operating the systems involved, it’s pretty much business as usual.
There’s another management console and a new set of terms to learn, but that’s not so different to any new system. Equally, it could be argued that availability risks increase if a number of physical machines are virtualised and run on a single server – but again, such challenges are familiar.
Even with these overheads, server consolidation via virtualisation does make operations easier. You’ve told us how you like the increased flexibility that virtualisation brings – being able to shift a VM from one server to another for example, so the service can continue uninterrupted even during scheduled maintenance.
So far, so good. Moving from this level up to ‘managed virtualisation’ causes more of a challenge, however. The more we look into this area, the less it appears to have much to do with the technologies involved, and more with the inertia of existing behaviours and practices. The majority of environments we research are what might be classed as ‘good enough’ – competently staffed and capable, but not looking to set the world alight as examples of best practice. While only a minority are spending their time fire fighting, a similarly small proportion is what might be considered ‘best in class’.
In principle, virtualisation brings with it the ability to respond to provisioning requirements quickly and efficiently. However, this also requires an IT organisation that has certain best practices in place, supported by appropriate tools and technologies.
Here’s the rub: while the majority of IT organisations are capable of implementing such things, most are unlikely to do so. This may sound like a sweeping statement, but for those who have been round the IT management block a few times, it’s a fair one.
Don’t get me wrong – this is not a slur on any IT departments or individuals within them. It’s more a question of priorities. Implementing operational best practice, tools and technologies comes at a cost, and like any initiative, it needs to be considered alongside other initiatives that may offer quicker or more obvious returns. You’ll know that the list of things you could do right now far exceeds the money you actually have available, so it stands to reason that something has to give.
It’s understandable, then, that recent research we conducted into virtualisation showed signs that activity was slowing down.
In particular, there are indications of a fissure between consolidation-type virtualisation, and ‘managed virtualisation’ of the form described here. If virtualisation is indeed a journey, it also makes sense that there is little evidence of the elusive level three of virtualisation, if the majority of organisations are baulking at the prospect of level two.
You have consistently told us that virtualisation brings benefits at every level, but the question of whether the benefits are sufficient to cause you to change how you operate is in your own hands. Many organisations may find themselves at a fork in the road when it comes to deciding what to do after server consolidation. We shall be watching with interest as we see whether the virtualisation continues, or whether organisations take the easier path towards other, lower-hanging fruit. ®
Sponsored: Benefits from the lessons learned in HPC