As we approach the holidays, many companies will be using virtualized servers to deal with extra traffic, but as more organizations embrace virtualization, it’s not just web traffic that is moving there, it’s also your critical apps.
And monitoring gets more complicated in a virtualized environment.That’s because as Sasha Gilenson points out in a post on CM Crossroads, virutalization is dynamic and most monitoring operations are static. As servers are added and removed, especially this time of year, it’s much more difficult to monitor what is literally a moving target.
And this is a particular problem in environments that have set up private cloud services with a menu of available resources. In the old model, you controlled the servers. If you needed a new one or a new license, you were responsible for bringing it online, loading the software and placing it under the control of your monitoring systems.
In the new model, you set up a portal where users can come and choose their own services. They might even be provisioning their own virtual servers through this system. They are certainly pulling licenses and using them for a given period of time.
One of the big advantages of this approach is that you can bill for what people use, and because they are paying for the services they use, chances are users will give them back when the project is over. Why pay for services you are no longer using, right?
From the end user perspective, they can begin to see the actual cost and value of IT services. When you are paying for every service you use, as with any company cost center, managers begin to understand in a very concrete way that these services cost money.
That’s all good for you from the perspective of people understanding the value of IT Operations, which as we’ve written is often difficult to communicate to the organization at large. But a virtualized service model makes life much more difficult for you when it comes to monitoring and tracking those services as people take projects on and off line — and you are left to monitor the resulting mess.
Yet even if you’re not offering self-provisioning, the whole virtualized approach to server and app deployment presents challenges to IT Operations just by its nature. As Gilenson pointed out in his post, IT Ops workflows are designed to deal with fixed environments and it’s going to take some rethinking in the core way you work to incorporate virtualized apps into the monitoring environment.
Have you changed the way you monitor because of virtualization at your company?