When is virtual machine technology the wrong choice?

When is virtual machine technology the wrong choice?

Summary: I was enjoying a conversation with the good folks at Appistry about where and how their application virtualization technology could best be used. I heard similar stories when speaking with the folks at Trigence.

SHARE:

I was enjoying a conversation with the good folks at Appistry about where and how their application virtualization technology could best be used. I heard similar stories when speaking with the folks at Trigence. I was impressed by some of ways their customers are making use of that technology.

During the conversation it again become clear at each of the various types of virtualization technology referenced in the Kusnetzky Group model of virtualization (see Sorting out the different layers of virtualization) has its unique place in the pantheon of information technology. Although virtual machine technology is just the ticket when system (client or server) optimization and consolidation are the goals, it simply isn't the correct choice when other requirements are at the forefront.

When is virtual machine software the wrong choice?

  • When the goal is allowing people to access applications and data from wherever they are, using whatever network-enabled device is handy and using a local network of some kind. This is when access virtualization is the best choice. By the way, the application and data being accessed could be hosted in a virtual machine on a local client, blade PC, blade computer or general purpose computer.
  • When the goal is highly efficient application isolation, application performance application reliability/availability or making an application work in an environment that normally would create problems. In this case application virtualization technology is a better choice.
  • When the goal is workload isolation and optimization, but all of the applications are designed for the same operating system. In this case, operating system virtualization/partioning is a better choice.

I could also bring up high availability/reliability, creating a unified management domain and a few other use cases.

Why is it, then, that organizations often fixate on virtual machine technology and try to apply it everywhere? I guess Abraham Maslow was right when he said "If the only tool you have is a hammer, you tend to see every problem as a nail."

Topics: Hardware, CXO, Cloud, Storage, Virtualization

About

Daniel Kusnetzky, a reformed software engineer and product manager, founded Kusnetzky Group LLC in 2006. He's literally written the book on virtualization and often comments on cloud computing, mobility and systems software. In his spare time, he's also the managing partner of Lux Sonus LLC, an investment firm.

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.

Talkback

5 comments
Log in or register to join the discussion
  • Message has been deleted.

    David Grober
  • Stop drinking the Kool Aid

    Virtual Machine technology is ONLY a "good" choice when you need to run server applications on "old" poorly designed M$ OpSystems i.e. pre-M$server2008. This is especially true for Exchange.

    When VMs are used with UNIX (including Linux), it creates an extra "layer" that saps performance. It also replaces automated provisioning services (memory management, CPU allocation) with a complicated manual methodology. How many sysadmins revisit IBM LPARS to re-tune performance after they are originally configured? No one I know. VMs also require more hardware (cost) to do the same thing (as without them) - Extra drives (mirrored) for booting each image, and extra network and HBA cards for each VM instance (not to mention the expense of a large, multi CPU frame).

    Higher cost, lower performance and replacing well tuned automation with manual configurations = stupidity = Virtualization/Consolidation.
    Roger Ramjet
    • VM tech is working great for us

      When we deploy a new app server, we put it on box X.
      Once that particular server needs more resources, we
      easily swap it to a dedicated box. If it doesn't, we
      leave it alone.

      It also allows us to give individual depts their "own"
      server without taking up as much rack space.
      jred
      • It would work better without it.

        [When we deploy a new app server, we put it on box X.
        Once that particular server needs more resources, we
        easily swap it to a dedicated box. If it doesn't, we
        leave it alone.]

        Ahh, but the opposite happens much more frequently - that a box X NEVER needs more resources - more like it's running at 10% capacity. You can either 1) Load a 2nd application on the server or 2) split the server using VMs and then add the 2nd application. What's the difference? I already spelled it out earlier.

        [It also allows us to give individual depts their "own"
        server without taking up as much rack space.]

        I have good news and bad news. The good news is that yes, I missed one (good) use for Virtualization - that of creating sandboxes for development. There are always ad hoc requests for a "box" for testing/development - and Virtualization is perfect for that. Of course that has nothing to do with Consolidation.

        The bad news is this - it is the individual departments that created the mess in the first place. Dept. A buys a huge server they don't need and their app runs at 10% utilization. Dept. B needs to run their app but they have no budget. If Dept. B sees the low utilization of the server and asks if they can use it - Dept. A tells them NO, It's OURS, get your OWN. So they do, and now there are 2 servers that run at low utilization. Now you see how it happens.

        With UNIX, it's a no-brainer to add as many applications to a single server as you want. Of course you have to keep track of the utilization - but isn't that what sysadmins are for?

        This whole virtualization scheme is an artifact of how finance/purchasing have failed at monetizing their purchases. They can only handle one Dept buying a single server, since everything is on paper and budgeted. They cannot handle utility computing (pay as you go), since the budget process is pay once per year.
        Roger Ramjet
        • Some agreement, but still

          If IT is considered a necessary evil, rather than a vital part of the business's core structure, then virtualization as Roger described it wouldn't be worth the headache.

          In my company, IT has a seat at the executive level. We have as much pull as Accounting, Marketing, and R&D. The other departments make requests to IT, and IT fulfills those requests in the most efficient and cost-effective manner possible, giving due consideration to the whole enterprise. We got away from that "my server, your server" crap. It's not Marketing's job to worry about how their data is stored, so long as they can get to their data.

          Now that we are in these tough economic times, server virtualization, thin clients and desktop elimination, and ubiquitous secure access, are receiving renewed interest.

          Although it may be the case that virtualization at this early stage is harder to manage, but that is what SysAdmins are for. Figure it out. Let me tell you, when the bonuses still flow because the company saved a bundle on infrastructure, I don't mind the extra work, nor do the users mind the performance hit. The technology will only get better, and ultimately, so will the bonuses for everybody.
          Worth2Cents