Virtual machines are wonderful things, but they can also cause serious headaches for network administrators.
That's because a fully configured virtual server can be created in minutes, and when it's that easy to create a server, that's what people will want to do. If an organization is not careful, it will quickly fall victim to virtual machine sprawl. Picture hundreds or thousands of virtual machines running on your network, consuming memory, disk space and CPU cycles, half of which you have lost track of or don't know why they were created. And when you do need to make or re-create a virtual machine, if you're not careful, you'll create one that does not comply with the regulations to which you must adhere, or or one that has costly license payment implications, or perhaps one that is subtly different from the one you created for the same purpose last time.
And that's one sure way of guaranteeing problems.
These, then, are the challenges that virtual machine lifecycle management software hopes to help you avoid. It's pretty clear that if you want to reap the maximum benefit from anything but the smallest of virtualized IT environments, you're going to need it.
At any rate, that's the position of the virtualization hypervisor vendors and many others involved in the business. There's certainly no shortage of companies offering software that promises to solve your problems. Microsoft is introducing Virtual Machine Manager 2007, which includes virtual machine lifecycle management, into its System Center family of products, VMware is offering VirtualCenter and Lab Manager, and Novell has Zenworks Orchestrator, VM Builder and VM Warehouse. There are plenty of others, too.
"Lifecycle management is about managing the sprawl of virtual machine images," said Larry Russon, a product manager at Novell. "You need to keep them compliant, which can be a problem, so knowing that people are checking out the right image is important. With a lifecycle management system you can know why people checked out an image, where it went, and any changes or updates made to it. You can check that it is compliant, and you know when it needs to be destroyed. Without it, you get in a situation when you simply don't know what all your virtual machines are for," he said.
Many lifecycle management systems work in similar ways. An organization either starts from scratch or sweeps its network to build an inventory of the virtual machines already in existence. It then builds a library or warehouse of virtual machine images and templates for different types of applications. This can be important for projects developed, tested and rolled out into production on virtual machines that must be identical or simply a way of ensuring new servers are compliant from the start.
These templates can also include self-contained environments, so a test lab can re-create a set of servers, clients and so on very quickly. VMware's Lab manager, for example, enables administrators to keep track of instances of an environment they have created. "That way, you can store a network, or send it to India for engineers over there to test and fault-find, for example," said Martin Niemer, a marketing manager at VMware. "You can control each environment you create, and apply a version to it," he says.
Most systems then have a system of "checking out" machines so their purpose, whereabouts, ownership and so on are recorded, and then checked back in when they are no longer needed. This enables them to be stored for later use if necessary, turned into templates for future machines, or simply destroyed.
In many ways, Microsoft has an advantage because its System Center family is concerned with running only in Microsoft environments. But what happens in complex environments where both physical and virtual machines run on a range of platforms?
"A true solution needs to manage physical and virtual machines, but it also needs to be able to look inside and manage the applications running on the virtual machines," said Dan Kusnetzky, principal analyst at Kusnetzky Group. "Applications may be split so that you have a database on one server, a presentation layer on another, and so on. What this means is that lifecycle management in the real world is actually much more complex than people initially expect. All of these management systems do something valuable, but none does everything," he said.
Perhaps this is not surprising. Because this sort of lifecycle management is still in its infancy, according to Niemer, companies are coming to grips with virtualization and the challenges it poses, as well as the opportunities it offers. "Right now, typical customers are getting into virtualization for server consolidation. The second thing they get into is disaster recovery, and the possibility of bringing up servers elsewhere very quickly. It's only then that they get in to the concept of lifecycle management. This process is only starting at the moment."
In other words, if you operate a virtualized environment and haven't heard much about virtual machine lifecycle management yet, you will soon. But don't expect it to solve all your problems just yet.