Change management in x86 server estates
Where's my @*>##% VM gone?
Workshop Changes to the IT environment are the source of some of the biggest headaches for the IT pro. They probably always will be, for the obvious reason that accidents aside, the most straightforward way of destabilising an IT system is to change it.
However, change is inevitable: by and large, dealing with such changes, and minimising their impact, is what the IT pro does for a living. When it's done proactively we call it change management, when it's done as a last resort, it's called fire fighting. The hard graft seen by many an IT vendor as a management software sales opportunity, is just another day at the office for countless numbers of IT operations personnel across the globe.
The situation may be getting worse, or be about to, if this were possible. The very point of technical advances such as virtualisation technology is ease of change. Many IT professionals might be forgiven for thinking that this is the last thing they actually want, or need. In environments where IT change management is a very manual task, using technology which can outpace the processes designed to govern it could be quite intimidating.
On the flip side, organisations in highly regulated industries which are obliged to run tightly controlled IT change management procedures may find dealing with virtualised server environments less of a big deal. However, even here, today's procedures may constitute too great an overhead for organisations wanting the benefits of more rapid provisioning through virtualisation.
Regardless, a couple of angles are worth thinking about in relation to IT change management processes. One is the fact that server virtualisation technology blurs the boundaries between previously discrete areas of the IT environment: namely the network, storage and server elements. This very notion offers the potential to improve IT change management simply by virtue of people having to talk to each other regularly instead of having “it's your problem” arguments at monthly meetings. Equally, we know from early adopters that getting people to talk to each other can be a major challenge.
The second is, given the very point of virtualisation is to streamline the IT department's ability to service the business, it would be nonsensical to constrain these abilities by forcing them to work at the rates previously used or expected of more traditional, static server environments. In other words, if you don't embrace new ways of managing the server estate, you risk only being able to exploit a small number of the benefits that could otherwise be gained. Not a good return for introducing an additional layer of complexity into the equation. Focus, then, needs shifting to the systems and tools in place, and the processes that govern them. Effective IT change management has always hinged on the ability to see and understand the impact of change across the IT estate (yes indeed Sherlock). Asset management and process automation software helps in this regard, but is not gracing the desktops of every systems administrator in IT today. As virtualisation technology is added to the IT department's portfolio, there is a risk of it being introduced and subsequently played with outside of existing provisioning and change processes. In order to keep track and keep control, many IT departments will have to mitigate risks by slowing down a bit until their management tools catch up.
On the systems management tools side, we are only just starting to see the management vendors acknowledging that really, it might make sense to be able to deliver a management view across ones IT estate, rather than in piecemeal chunks. While things are slowly moving in the right direction, it remains unlikely that a single management vendor will ever be able to deliver a one stop solution for Holistic Datacentre Lifecycle Management. Not least because the individual elements of IT are just too darned complicated for any one tool to manage.
However, if the management of your recently virtualised servers is currently an addendum all the other management stuff you do, it might be worth considering how to bring it into the fold. Process needs to come first, because ultimately it's where the discipline and consistency lives. Next come tools. Fortunately some of the vendors you undoubtedly already spent money with are starting to get the drift and are able provide tools to help manage heterogeneous IT estates. The open source community is also quite alive in this space.
Of course, an easier option is to tell us all about how you currently manage change across your heterogeneous server estate, and why making special allowances for the virtualised parts does, or doesn't matter. ®
Sponsored: Hyper-scale data management