An interesting (to me) lab conversation

We are in the process of upgrading a file server we have which we use to store DMP files. We’ve had this machine for about 6 years now and it was starting to get full and remain full. As a file server we were only really interested in storage capacity when we bought it so it had only a single core Pentium 4 but plenty of extra hard drive bays. Over time, we had maximized its capacity and we noticed that the machine was running slowly by today’s standards. So we ordered a new blade server and needed to go to the lab to get it all set up.

While we were in the lab, one of the lab workers was there to help out with naming the new machine. As we were waiting for our changes to propagate, we looked at one of the desktop machines that we in the lab. It had a Windows 95 sticker on it. That’s a pretty old machine to still be in use. There was also a machine that had a Pentium 2 sticker on it and a few others that looked to be that old as well.

The lab guy wanted to virtualize all these machines for several reasons. He wanted to save electricity (I’m all for that), space in the lab (again, I’m on his side) and he mentioned that the virtual machines would run faster than the old hardware (I want to see numbers on this first :)).

But overall I disagreed with him. My point of view was to leave the machines as is since they are not broken and are working just fine. Moving them to VMs would take time, energy and have the potential to break. At that point, not only would I not be better off, I would now have to devote even more time to rebuild whatever was broken in the process. To me, it is easier to just leave them working as is. If they break or otherwise need to be replaced due to hardware failure, then I would easily be convinced to move them to a virtual machine.

It is kind of interesting to me to see these two different points of view. Both are solid and valid viewpoints from two different engineers looking at the same problem from two different vantage points.

And as it was, no one had time to convert the machines, so we left them as is.

Questions, comments, concerns and criticisms always welcome,

Comments (4)

  1. Josh Smith says:

    A classic case of "just because it -can- be done doesn't mean it -should- be done."

    Or "if it ain't broke don't fix it." 😛

  2. John says:

    That's a big 10-4 good buddy.  

  3. Paul Cavanagh says:

    Why not leave the old h/w as is, but snapshot them so the lab engineer CAN prove the snapshots can run in the virtualised environment, then you have a fallback for when he old h/w fails….and they can prove to you that the vm will run faster….. or maybe not

  4. John says:

    Well, snapshotting the existing machine would take a good bit of time since I would need to find some old archiving software and then find time to run the utility.  I just don't have the time to spend right now.

    Now, if (and when) the machine fails, we can go through all this to get it restored.  But until then, I don't want to spend the time trying to fix something that is currently working.

    Make sense?  It's not rocket science, and most tech geeks I know ALWAYS want to fiddle with (improve) whatever is at hand, and that is a good thing.  But I would rather spend the time improving OneNote than moving machines around in a lab at this point in time.