When organizations put a lot of eggs in one basket - desktop side of the story

When organizations put a lot of eggs in one basket - desktop side of the story

Summary: A while back, I posted about some of the availability/reliability problems that can arise when many workloads are hosted on the same physical server (see When organizations put a lot of eggs in one basket for more information.) Several readers pointed out that there is more to that story.

SHARE:
TOPICS: Hardware
6

A while back, I posted about some of the availability/reliability problems that can arise when many workloads are hosted on the same physical server (see When organizations put a lot of eggs in one basket for more information.) Several readers pointed out that there is more to that story. What happens when desktop workloads are encapsulated and run somewhere in the network? As one might expect, similar problems of availability and reliability can arise.

Let's look at different possibilities:

  • Workloads could be encapsulated using application virtualization technology and delivered to a remote system (could be a netbook, a laptop, a tablet or a desktop system). If the physical machine fails, suppliers such as AppZero, Citrix, MokaFive, Microsoft Spoon and VMware would suggest just sending the application back to the system once it is back on the air. Some also offer the technology necessary to replace user's data as well.
  • Multiple virtual desktops hosted on a single PC or desktop system — products from companies such as Citrix, Neocleus, Virtual Computer and VMware allow individual workloads to be encapsultaed into virtual machines and then many of them to be hosted on the same physical laptop or desktop computer. If that physical system fails, all of the workloads and the data they were processing can come tumbling down. Several of the suppliers point out that it is simple to just reload the workload from a remote or local server. They don't have an easy answer to the question "what happens to the data being processed at the moment of failure?"
  • Multiple virtual desktops hosted on a single blade PC — products from companies such as Citrix, HP, NComputing, Microsoft, Pano Logic, Wyse, and others individual workloads to be either run directly on a blade PC or encapsulated in virtual machines. If that blade PC  fails, all of the workloads and the data it was processing can wind up on the floor. come tumbling down. As in the case in which the physical system was a single PC or desktop system, reloading the workload can be relatively simple. It is not always as easy to return to the exact state prior to the failure.
  • Multiple virtual desktops hosted back in the datacenter or in the cloud somewhere — the same problems mentioned before can occur. Several of the same suppliers Citrix, Microsoft and VMware as well as a few newcomers, such as Skytap, suggest that it is easy to deal with such failures ... if the organization is using their software suite.

In the end, availability and reliability need to be a central part of the planning for putting all of the organization's desktop workloads in a small number of computing baskets.

Topic: Hardware

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

6 comments
Log in or register to join the discussion
  • Think of a RAID array....

    When one virtual system goes down, there space in the server room for the backup to kick in. Or in this case, the Parallel virtual machine to take over. I think you have it all wrong and not thinking how a Datacenter works.

    To put it straight forward, companies who setup Virtual servers also have their backups, those who don't fail in the matter you metioned. Now the backup system is normally running in parallel with the other so if something does happen, the incoming traffic will go to the other server. The same way a SAN or RAID array works. When one HD fails the data on the other takes over without any loss of data. The result in the server is hardware being replaced and being brought back online without any interruption. The only thing users will see is a bit of a slow down due to the lack of processing time from the remaining CPUs.
    Maarek
  • Look into Single System Image (SSI)

    There are a number of products/projects offering SSI functionality, which is essentially the opposite of multiple virtual machines on a single real machine, and more like the RAID mentioned earlier: Multiple real machines are ganged together in a way which appears to the applications to be a single system running a single OS. Each node contributes all its resources to the community. Any node can fail, and the other nodes pick up the slack.

    I'm pretty sure Google uses something like this, and that's why their custom-built blades even include hard-drives and battery backup; because they're more than mere blades: They're complete PCs (sans screen and keyboard), meant to be a fully operational node, contributing not just CPU and RAM, but HD and power to the whole.

    I think it's brilliant. And it's about time, too. Can you just picture home or small business servers made of a small, user-expandable stack of interconnecting modules accessed via thin-clients or phone apps?
    Gaius_Maximus
  • on a purely mathematical level, they are the same

    Let's say I have 100 desktops providing 99% uptime. That means that for any given hundred days, there will always be a broken desktop.

    Now, let's put the same 100 desktops virtualized on one server, and for the sake of argument, the SAME 99% uptime. So, once every 100 days, the server goes down.

    In reality, stand-alone servers are substantially more reliable than desktops. Three and four "9s" are common, factor in simple clustering of two machines, and 99.99% uptime is affordable. Then factor in desk side repair, adds-moves-changes, electricity (120w for a desktop vs 12w for a thin client) and the pros for virtualized environments become unassailable.
    caspianhiro
  • Agreed

    Putting all of your eggs in one basket is not a good idea, agreed. Currently I am contracting for CloudShare and let me tell you with the right planing, implementation and tools there is no reason to worry about failure especially with CloudShare (www.CloudShare.com). I have used CouldShare Pro personally for virtualizing a Windows 2008 server with AD, and it?s flawless! You can check out more about the IT operations part here: http://www.cloudshare.com/solutions/ForITOps.aspx#ReduceIT. Also, you can use CloudShare for free (http://www.cloudshare.com/Products/CloudShare-Pro/CloudShare-Pro-overview.aspx), unlike Skytap as mentioned.

    Thanks for the post!
    TheGerm
  • HP announces enhanced desktop virtualization offerings at Citrix's Synergy

    Where does Idaho rank? We have been living in Montana for the past 5 years and I am not supri<a href=http://www.hipersexshop.com.br>sexy shop</a>to find it #3 on the "worst" list. Considering a<a href=http://www.hipersexshop.com.br>sexshop</a>move to Idaho to escapthe high cost of living a low income in MT. There may not be a sales tax here but they get you if you own property!
    jeffmgf1
    • RE: When organizations put a lot of eggs in one basket - desktop side of the story

      @jeffmgf1 are u want to buy <a href="http://www.nikefreesalecheap.com">Nike Free Run Sale</a> shoes? visit <a href="http://www.nikefreesalecheap.com">Nike Free Run Cheap</a> then
      zdne110