As cloud computing, or the use of remote, Internet-based computing services, continues to grope its way through infancy, we can anticipate that an initial area of success will be in virtual data centers.
Hang on, keep reading: I am not talking about outsourcing your data center. The initial incarnations of cloud computing will be from companies like Skytap Inc. and Q-layer offering virtual labs or training centers over the Internet.
Virtual labs have two key roles that can be utilized safely today. The first is using the service to extend your current lab. With this type of technology, you can “dial in” the server resources you need as you need them.
Say you wanted to test a new Linux-based database and needed 10 Linux systems to run the database and 10 Windows systems to act as a client accessing the database. In a physical world, this could be a costly and time-consuming event on your part. Reality would typically interfere. You may not actually have the time or resources to configure a test suite that large, resulting in a test that does not properly gauge the new application and may leave you exposed to bugs when you move into production.
With a virtual lab, you would use a Web browser to design the system requirements. The service provider would create all the virtual machines that you need in the background.
You can also use virtual labs as an extension of your current lab. By providing a VPN connection, you can allow the virtual part of your lab direct access to your local physical lab. This would have value from a final-phase testing perspective, where you want to scale up the number of users or servers to stress the new application.
The second area of interest for cloud services is for training centers. Instead of shipping expensive hardware around the country or hoping that a local training center has the appropriate hardware, you could send your preconfigured virtual machines to the virtual lab, and then when you’re at that day’s training location, all you need is a basic desktop with Internet access to present the appropriate server or workstation to the user being trained.
This should result in a more consistent and stable training experience for the students, since the physical systems never really move. It also makes it easy to re-image the lab back to its pre-class state to be read for the next class. Either situation brings flexibility and agility. If there is a need to pause testing for a period of time, you won’t incur the lost opportunity cost of hardware sitting idle. Simply save the environment until you are ready to come back to it. In training, if an extra student shows up, copy one of the current profiles to a new virtual machine, provision it, and you are ready to go, no scrambling to get additional hardware or software.
There are downsides to virtual labs: One is the obvious immaturity of all things cloud. Second, virtual labs are mostly for Intel environments today. Third, you can't furnish the service network -- the cloud -- with hardware; that's an element of control that you will always lack.
Still, virtual labs provide a safe and legitimate entrée into cloud computing, while at the same time offering a significant cost savings in an area that is hard to fund -- and, more importantly, hard to manage.
— George Crump, President, Storage Switzerland