Commentary--If you're like most IT managers, your boss probably expect you to do more with less: expand your IT capabilities and infrastructure while doing it with reduced funds, staff and hardware. It almost sounds like a paradox, but emerging technologies such as virtualization make this possible, allowing you to you can do more with less. It is a question of knowing how to put the pieces of the puzzle together. Virtualization requires a new way of looking at technology and how it can work for you. There are four kinds of virtualization: server, software, storage and network. This article examines,examines server virtualization and migration issues.
Virtualization is not new. In fact, you have been doing it for years without really thinking about it. Nearly every IT manager and consumer is familiar with the most basic form of virtualization: partitions on a hard disk. When you create a partition and assign it a drive letter, Windows sees that partition as a virtual hard disk. As far as Windows is concerned, it is a separate, physical device. Only the software knows that this drive is really just part of a single physical drive that goes by multiple names.
Now consider: What if that virtual hard disk had its own operating system and applications? In fact, what if you had multiple partitions and each one of them had its own operating system and applications. You could effectively run several computers with one CPU, motherboard, video card, network card and disk drive. That is the underlying logic and business case for hardware virtualization.
Why virtualize?
There are many reasons why an IT department might move from a traditional, physical server-based infrastructure to one based on virtual technology. First and foremost is, of course, money. Whether it’s the cost of cooling thousands of physical servers, powering the boxes or simply finding the real estate for them, a purely physical server infrastructure can be expensive. It all comes back to doing more with less.
Virtual servers are an excellent choice for consolidating servers and reducing indirect computer-specific IT costs. Prototyping and software deployment also are improved dramatically using virtual technology. The challenge, however, is understating how and why the dichotomy of less is more--or physical to virtual (P2V)--works so well in computer processing.
Another reason to move to virtual servers is systems management. Keeping track of thousands of servers can be a management nightmare; managing just a few or a few dozen can turn that nightmare into a viable operation for a scaled-down IT department.
Once you’ve made the move to the virtual world, your infrastructure will never look the same again. Not only can move processing services from physical to virtual machines, but you also can move them from virtual to other virtual (V2V) machines.
Finally, the operation comes full circle when you move processes from virtual to physical machines (V2P) or physical to physical machines (P2P).
What, you ask? Why step backwards and move to a physical machine when virtual systems are fast and easy to use? We’ll see in a moment.
Physical to virtual
Virtualization is used for prototypes and development, as well as for production systems. Prototyping is often required when companies are developing new applications and testing them on various hardware configurations. Unlike physical servers, which might require a full OS reinstallation and reconfiguration if a software package crashes, virtual servers can be recreated in minutes. As a result, this approach is a very time- and cost-effective approach for software development and systems configuration.
An important benefit to virtualizing hardware on production servers is the ability to do more with less — create more computers without buying new hardware; run more servers without increasing your capital or real estate investment to house the additional boxes; and increase processing power without spending money on more electricity or air conditioning. In fact, you often spend less.
The beauty of this approach is that with multiple servers running various applications simultaneously, you get all of the benefits of a server farm without the management headache. The challenge, however, is making sure you have enough horsepower to run all of these disparate servers as well as having the appropriate hardware for each.
Virtual to virtual
As noted earlier, software prototyping is a popular use for virtual systems. Often, these prototyped systems will then be deployed to other virtual servers. When this happens, you need a simple way to ensure that these servers will run on the new hardware.
Ideally, your deployment tool will be able to recognize the various virtual environments, whether it’s from Microsoft Virtual Server, VMware, Parallels, Xensoft or similar product. Then, when the deployment is done, the tool will make any necessary modifications automatically and without user interaction.
Another way to accomplish this is to use a deployment tool that permits you to restore the image of the server to dissimilar hardware. Such a universal restore tool needs to offer a centralized driver repository as well as the ability to install your own device drivers. Make sure that when you are buying such software, you get an application that is integrated with your disk imaging software. That is the best way to ensure you can deploy images to any virtual server from any virtual server.
Virtual to physical
With all the discussion about deploying virtual servers, it seems almost contradictory to discuss deploying physical servers from virtual ones. However, this is a very real possibility when you consider that there are some applications, such as those that require significant processor cycles, that simply will not work in a virtual environment.
For example, let’s say you start out with a small database that has a few hundred or up to thousands of data points. Re-indexing such a database might not require significant CPU resources. Now let’s say you’ve grown that database to millions of data points. Now when you re-index, your system could come to a grinding halt as applications vie for CPU cycles.
In cases where software is unable to run on a virtual machine, you need a mechanism to move back to a physical environment. While virtualization vendors are happy to provide tools for moving from the physical to the virtual environment, they tend not to offer tools to move from the virtual back to the physical world. Indeed, none of the virtual OS providers offer tools to move from one virtual OS to another. However, moving from a virtual to a physical server is an all-too-real possibility and it is critical that you have tools in place before you begin your deployment so that if you find you need to move back to a physical server, you can do so swiftly and painlessly. Similarly, you might well need a tool to move from one virtual environment to another.
Again, we’ve touched on just a few issues that deal with server-based virtualization. Employing virtual technology requires more than just a shift in how you use servers, but also how you think of your server. Using virtualization rather than strictly a physical infrastructure can provide you with a vast array of hardware possibilities. But with each possibility there is a corresponding challenge. Make sure that when you are ready to start testing and deploying virtual servers, you have a viable image of that server that you can restore.
Remember, just having a backup of a virtual server is meaningless if you can’t restore it. A powerful imaging software application, combined with the ability to restore that image to dissimilar hardware, will do wonders for your confidence and systems reliability.
biography
Acronis CEO Walter Scott has over fifteen years of software experience. Prior to Acronis Walter was President and CEO of Imceda Software, a database backup software company which was sold to Quest Software in May of 2005.
