Virtualization Technology Adoption – part 3 – Desktop Computers
It’s Wednesday, the middle of the week, and also the middle of this series on analyzing virtualization technology by market adoption and value proposition. We’ve covered mainframe computing and compute servers, equipment the average person never sees, and even the IT professional usually administers remotely. We all use this type of computing, when we see a YouTube video, its likely streamed through a compute server, and when we perform a bank transaction, the host is likely a mainframe. To quickly recap, the mature mainframe market would see utilization of all the enumerated value propositions; the compute server market saw the growth of VM adoption driven largely by the first three values (dark green boxes) and will continue to see further adoption and enabling cloud computing through the values of portability, deployment and security (light green boxes).
Virtualization Value Proposition by Computing Class
When it comes to desktop computing, we are dealing with a technology that is more hands on for everyone. It is interesting to note that virtual machines are not new to the desktop, for the market leader VMware, their first product was desktop virtual machine software. It allowed your desktop machine to host an additional operating system, even a different operating system than the one running on the host machine. It allowed a Windows machine to run Linux in a virtual machine. There were also vendors who provided virtual machines for the Mac OS so that a Windows OS could be hosted to run Windows only applications. In this case the primary value proposition is heterogeneity, the ability for one computer to host multiple operating system environments. This use of virtualization also proved useful as a test environment, where new software could be developed and tested in a virtual machine. If the software failed catastrophically, it would not corrupt the host machine, the virtual machine could be scrapped and a new one instantiated for the next test. This is the isolation value proposition.
At this point it is worth noting that virtual machines on a desktop computer are not the same thing as desktop virtualization. In the former, the desktop is the host, whereas in the latter a server is the host of the desktop environment. The latter case also has different value propositions. In this case it is deployment and security (the light green boxes on the chart); the federal government deploys desktop virtualization to employees and contractors primarily for these value propositions.
Technology Adoption Lifecycle – part 3
Although virtual machines have been on the desktop at least as long as on compute servers, their deployment is not as wide spread. Thus I would place it earlier on the Early Majority part of the Technology Adoption Lifecycle curve. Now that multi-core processors are not only common on desktop computers, but also on laptops, I would expect the adoption of virtual machines on the desktop to continue to grow, but I do not expect that it will catch up with virtual machines on compute servers.
Next we’ll look at embedded computing.
April 29, 2010 - 10:14 am
I’ve used virtualization on a desktop for testing purposes. It’s a useful way to reset or to revert back an environment.
VMWare has really improved its product, especially with GPU acceleration support and support for USB.
One thing that gets to be difficult in managing is backups. VM’s tend to get very large, and if you have many VM’s that is a lot of backup work. This is a limiting factor for widespread adoption of VM’s on a desktop and most especially a laptop. For example, maybe it’s better to host VM on a desktop that has RAID 1 or 5, instead of on a laptop, which runs on a single drive.
If the drive goes bad the user loses everything (and running regular backups for VM’s becomes time consuming).
Do you think something as simple as USB 3.0 would speed up adoption?
April 29, 2010 - 6:57 pm
Thanks for your comments Chris, you have brought up additional considerations worth noting. From my perspective, back-up was simplified because everything is encapsulated in the file. It was easier for me because I was working with diskless servers, and I was using NetApp utilities to replicate my data.
I comment on desktop computing, as I am applying my thinking about value propositions and market adoption for different classes of computing, I’m not claiming to have equal experience in all these environments.
I recognize these files can get very big and when there are many of them, a desktop computer and more so a laptop is not designed to handle that kind of data replication. It makes sense that the data to back up is multiplied, as VMs multiply the number of computing environments supported by a host. Perhaps this will be an incentive to migrate from VMs on desktops to desktop virtualization. From a data protection perspective, I really think you want replication outside of the machine, so maybe the answer is faster external connections (like USB 3.0) and better desktop back-up solutions.
When I did my work, backing up running VMs was not possible as far as I can recall, if we could do that, back-ups could be done in the background.
I’m also a fan of solid state drives. Although they are more expensive and less dense than magnetic drives, they are faster and more reliable. I would think that larger laptops would have room for duplicated solid state drives, and in this case fast replication of VMs could be supported internally.
It should be noted that for compute servers also, adding virtual machines increases the requirements on storage and IO. Not surprising then that a high end storage vendor bought VMware, eh?