(Excerpt from original post on the Taneja Group News Blog)
A great recent trend in virtualization management is to intelligently integrate “analytical” functions like capacity planning with active operational processes like remediation and provisioning. Each individual management activity has had its challenges with virtualization – capacity planning has had to learn to pierce through layers of abstraction to piece together the actual infrastructure components in play, while operationally doing anything smartly requires a thorough grasp of the dynamics built-in to the virtual management layers (e.g. VMware DRS and Storage vMotion). But as these individual management capabilities mature, the next level of value comes in leveraging them together to make smarter, more automated environments.
When Quest acquired VKernel to augment and extend their (v)Foglight solutions, it was probably thinking about this higher level of intelligent automation in the virtualization space. After all, virtual admins have got quite a lot on their plate, and as more and more mission-critical apps virtualize, multi-tool management operations become onerous and error-prone. For example, the latest vOps helps its admin users see historical configuration changes on a timeline perspective against performance metrics, review a ranked list of changes by potential risk, and to revert or rollback each change if desired. You could compare this to the latest vCenter Enterprise edition which also enables charting and rollback of configuration changes but at a higher price and without the risk evaluation. VKernel’s vOps also has an existing one-click feature that can add automatically identified critical resources to constrained vm’s (e.g. “add a CPU” button shows up when a vm is compute constrained) to accelerate remediation in support of tight SLA’s.
On the planning side vOps had previously enabled admins to set hard reservations of resources for future vm deployments based on an identified vm template. In large multiple administrator environments this helps ensure the right resources are going to be available on day 0 for new vms. They’ve now enhanced their active provisioning so that it initially deploys the right vm’s into their specific reservations in one “atomic” step, avoiding having to either first manually release the reservations or temporarily over-subscribe the system. Remember that virtual systems are dynamic, so releasing reservations manually ahead of deployment can cause other things to inefficiently “shift” around. And manually keeping track of reservations mapped to deployments is likely to lead to orphaned reservations floating around. You definitely don’t want reservation “leakage” to add to your vm sprawl problems!
Note that the virtual admin is still in the loop on these operations tasks, but the upfront analytical “expertise” is getting baked in. Fully automated remediation and performance-based provisioning are still in our future, but we suspect those capabilities are eventually going to become the ultimate definition and real value of “private cloud.”
…(read the full post)