Windows Server 2012 means re-evaluating our OS support matrix

I was speaking with a server support manager that I had met for coffee over the weekend to ask him what he was working on, the goal of the blog has always been to keep me up to date, to learn and share knowledge. The conversation has been anomalized, by that I mean, I have removed anything that might identify the contributor or their employer, the content as always is just as relevant for your review.

So what are you working on?

At the moment, we’re just completing the final elements of our data center consolidation project, it’s been rather busy, it’s consolidating our data center environments to reduce cost, with tight deadlines from a business and a technical perspective at the same time the server teams were examining what routes to take. In some cases, we might build a new server and migrate the application or workload, in others it might be that we virtualized and in a minority of cases where timelines or complexities could introduce delays we did a lift and shift, (physically move the server to the new data center).

What challenges did you have?

The main ones we had were in managing expectations (everyone wanted either their systems done first or last depending on their environment and faith in the process). At the same time, not everyone appreciates the logistics involved in a data center move, it’s not just the physical or virtual move of the server, there are the pre-checks to verify (particualrly if there are IP changes involved) as well as the implementation steps to encompass the infrastructure, application teams and user community. Also there were times where we identified systems or components that were not working as expected or not configured as we thought they were, the ILOs were in DNS but the board not responding or not being configured correctly, it’s only therefore when you try to access it, move it or reconfigure it that these issues were identified. In the application space during testing we encountered a few issues where the server was moved, the application tested only to identify that a configuration or setting was not in place causing delays and resulting in further testing.

What are you working on now?

There is a concerted effort to start testing Windows Server 2012 in our environment, to have the server build and architecture team to investigate an alpha or beta build so that we can validate the build, test in our environment on our hardware stack with our layered products in preparation to meet any requirements for Windows Server 2012.

The onboarding of Windows Server 2012 as a supported or ‘to be supported’ platform raises the issues of what we should be doing with Windows Server 2003 and early Windows Server 2008 (non R2 builds), should we be only decommissioning Windows Server 2003 servers, or should we standardize on Windows Server 2008 R2 as the platform going forward and migrate applications and upgrade servers to this OS or Windows Server 2012 as part of consolidation of roles, reducing costs and keeping the server platform moving forward and up to date with what the developers are after.

We’re also discussing a few procedures and best practice discussions, we have been trying to make x86 (32bit) operating systems unsupported or delivered by exception and in the majority of cases this has been working well, the next objectives is to evaluate standards and manage that difficult line of reliability or industrial strength with cost, cost vs reliability so to speak. An example, do we need network teaming, do we use the on-board network cards or separate ones so that we can deliminate between a system and network problem, (the onboard network ports using the same path could be seen as a central point of failure).

The next thing I’m trying to manage is the questions I’m getting about the cost of the hardware support contract which covers all servers we support, specifically do we need a separate support contract or could we cope with a 4 hour response with some spare servers or gold stock to back fill any scenarios where it might not be a hardware failure but we need to rule it out.

This site uses Akismet to reduce spam. Learn how your comment data is processed.