Virtualization of legacy environments still makes sense

NT4 web site

I was having a chat with a manager that I had met at one of the conferences I had attended in the last few months, anyway he was telling me his challenges during one of the debates about virtualization and next generation technologies, the conversation is below, I have removed comments identifying his organization, his statements are indented.

“That’s great what he’s saying but that’s not my problem, my problem is the 37 Windows NT4 servers running on obsolete hardware, what do I do with these?”

Virtualize them, it’s easy, there is no reason that you can’t take your old Compaq/Dell or IBMs and virtualize them, you can even as part of this increase the disk space and the memory, making them a bit more supportable. I replied.

But NT4 is out of support, that does not solve the problem I have supporting NT4 servers.

No it does not, but it solves the main one, and let us not forget that:
1.) We as end users are always going to be out of support, the vendors have their recommendations and best practice, but I live in the real world, where my application has been coded for NT4 cannot be easily ported to Windows 2000/2003, in which case, I would rather have an easy to support virtual NT4 server than worrying about how the server was built (all raid5 etc), whether the server will come back if I reboot it or that it only has a 2GB c drive which is constantly full and generating alerts/workload.
2.) Hardware is relatively more expensive to support operationally and in terms of power, in which case I would rather have something that is out of support on a more manageable hardware platform.

My best example, the small business I met last week, they had a Compaq Proliant 1850 with three disk shelves attached to the back of it, proudly giving them 400GB of storage which had been carved up into various drives for different business units. Their concern? That the desktops and their small log onscripts etc all pointed to a certain name and no one wanted to change that, but the server would not run Windows 2008, so they were stuck.

That Proliant 1850 could easily be replaced with a 1u box running VMware or the equivalent with a couple of large drives, saving power and the support costs in terms of the older hardware. I said.

But you are still running NT4. – what do you do after you have virtualized? Aren’t you taking your existing problems into your new clean virtual environment?

Well you could say that, but then I have always preferred to be making some progress rather than none. I need to reduce my hardware support costs mainly from a marginal cost base standpoint, I do not want to be spending company money and effort supporting 4GB/9GB disks in old servers which are using more units of power per performance than I need to. 

We cannot change the universe overnight, let us make small steps to make life a little bit easier, with the older hardware gone and out of the debate, I can then focus on understanding what the barriers to moving off NT4 might be, what we can do to resolve this going forward.

Running legacy platforms is only a problem, an issue if you make it so, if your systems are configured, if you have the right systems in place, being in a position to support (granted possibly on a best endeavours basis) might not be that difficult or represent that much more risk over a newer platform. That the business unit, the sponsor understands the associated risks of their platform, might mitigate those risks, the is it £2 million we want to pay for a recode of an application or service which is going next year, or £100,000 we would rather spend on virtualization to extend its life and reduce support costs until we replace or decommission it.

Added to this, there is no reason that as we virtualize each of your servers that we cannot clean them, right size them as we go, we can make the C drive that little bit bigger, 4gb instead of 2GB, give that d drive for the logs a bit more space than the 4GB or so, and we can give it a bit more RAM 1GB instead of the 256 or 512 it has now, all these things can improve performance and allow us to make mini improvements at marginal cost.

No Responses to “Virtualization of legacy environments still makes sense”

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