Orchestration: Virtual Server Process Management

Many of the oft-touted benefits of moving to a fully virtualized data center revolve around automation: provisioning new virtual machines, on-demand services, succinct use of resources, etc. When discussing virtual automation benefits, the anecdotal saying “Virtualization allows IT to spin up a new server in six minutes instead of 60 days” is one that’s rooted in experience and truth.

When you dig deeper into why it was taking IT 60 days to turn up a new server, however, you’ll immediately find that the reason it took so long was mostly due to a very long list of processes: justifying the business need, spec’ing the server, ordering the server (making sure it meets the current IT standards, another list of processes in itself), installing the server, loading the base image, so on and so forth. In other words, that 60 days was an IT constant because there were a lot of steps that couldn’t be glossed over.

Virtualization of servers changes almost all of those steps and actually removes a good portion of them, but not without cost. Just as those 60 days for a server install were dependent on a process, virtualization’s six minutes for a virtual server are feasible due to an almost complete lack of processes. In fact one of the very first concerns with virtualization came from this lack of process management: virtualization sprawl. Simply introducing virtual servers into the data center (and anywhere else within the organization) can actually start breaking IT processes that have been in place for years.

That’s not to say that virtualization itself is causing a failure in process management. One of my favorite IT sayings is “Don’t blame the technology, blame the implementation,” which comes from years of worrying about who to blame during an incidence response post-mortem. More often than not, how the technology is being used and deployed is causing the failure more than the technology itself; such is the case with virtualization and process management. Virtualization isn’t to blame for virtual sprawl, it’s the lack of process management in most virtual deployments that’s to blame.

It’s not all doom and gloom, however. Virtual platform providers have been on the forefront of releasing management tools in step with their virtual solutions. They recognize that virtual platforms are basically complete data centers in a box: every tool we use in a physical data center can now be deployed virtually, including extremely sophisticated management tasks, typically dubbed orchestration solutions. All three major virtual platform providers ship with their own management and orchestration solutions: VMware with vCenter and Orchestrator; Microsoft (News - Alert) with System Center and Virtual Machine Manager; Citrix with Essentials. Each of these tools offers standard virtual machine and resource management as well as more advanced process management and orchestration. It’s not enough to simply know when a virtual machine is up or down; these tools must know everything about the resources required to run virtual machines, where those virtual machines are currently running, and probably most importantly what those virtual machines are doing. Virtual orchestration solutions are both the brains and the brawn of a virtualized data center.

Even though all the major virtual platform vendors understand this and are doing a great job at looking beyond simple management with orchestration tools, one of the concerns of a virtualized data center will be vendor lock-in. Competition for turf from the big three virtual platform providers includes both the hypervisor platform and now orchestration tools. Much of the decision process on choosing a standardized virtual platform now includes how to manage and orchestrate those platforms, and the market is already starting to align specific orchestration needs with specific deployment types.

That said, all three vendors are very actively pursuing third-party integration within each of their management platforms. They do understand that most virtual deployments will contain a mix of systems, hypervisors, and virtualized application workloads, and they each need to offer solutions that work with existing management and orchestration tools.

Maybe this openness to integration is simply a reflection of market drivers to make everything openly manageable with Web-based APIs; regardless of the reason, integration is a positive move for orchestration and virtual machine process management.

Virtualization’s popularity and massive adoption rates are due in large part to the ease of deployment with virtual machines and how quickly IT is able to move from the 60-day model to the six-minute model. But that same agility also is still a major concern with virtual deployments, and the more IT relies on virtualization the more of an issue management and orchestration become within the data center. Managing a virtualized process that will eventually include services beyond the data center and in the cloud…that’s a conversation worth saving for another column. IT

Alan Murphy is technical marketing manager of management and virtualization solutions with F5 Networks (News - Alert) (www.f5.com).

» Internet Telephony Magazine Table of Contents