Private Cloud Orchestration: A Cog in the Wheel Part 2

Don’t Stop Believin’

Those of us who are frighteningly optimistic know that “challenge” and “opportunity” are synonyms. In this case, how do we address this abrupt handoff of the work of an private cloud orchestration engine to the infrastructure? First, we have to stop treating cloud orchestration and management as two separate processes because they really are not. The control of a workload lifecycle, from orchestration to decommission, is one continuous process. Granted there are a huge number of factors that contribute, but the goal of this concerted effort is to deliver a service. We need to think this way if we are going to marry the orchestration process with provisioning and management.

Furthermore, this endeavor must be holistic. In order for a self-service organization to be successful, the walls between the aforementioned IT teams must cease to exist. After all, what good is capacity if we cannot manage it? What good is flash storage if we do not have the compute resource availability to compliment it? Four-socket, 1TB hosts are worthless without a network that is up to snuff. IT organizations must function the way their virtualized infrastructures do: share everything (yes, including responsibility). In this vein, we must integrate orchestration into the rest of IT, not separate it into its own little box.

This means that private cloud orchestration must reserve capacity when appropriate. If we have foresight into a project and understand its requirements, we should carve out the necessary capacity before the actual orchestration and deployment to guarantee its success. Upon successful orchestration, we must have the ability to deploy in the best possible part of the infrastructure to ensure the application in question may perform as expected from day-zero (and, naturally, this cannot be accomplished without the aforementioned reservation of capacity). Lastly, the platform that centralizes and controls this infrastructure must be driving these sorts of decisions at the orchestration layer.

Sequence: An Enemy in Disguise

By now it should be apparent that orchestration solves a serious challenge but also presents (and even perpetuate) others. I do not mean to imply that any organization, however, large or small, should be averse to adopting private cloud orchestration and self-service—quite the contrary. Rather, orchestration is an element in a broader overall shift in the way IT operates. One of the major concerns I personally have with orchestration is that, if we do not change the way we think about IT, then orchestration will not have the positive impact we desire or anticipate. Orchestration is not a solution for a sequential IT organization; i.e., if IT thinks of service delivery as user à request à orchestrate à provision à manage à decommission, then the benefits of orchestration are relatively minimal.

Think of these processes as a web rather than a sequence. A web, implying that each element is aware and may even initiate, impact, or control any or every other element. After all, if our application’s service or response time is not satisfactory, what should be done? Should a user be required to request another instance of that application, or should the control platform automatically understand an increase in demand from the application in question, evaluate , and then execute the correct decision to accommodate that increase while still maintaining the desired SLA (e.g., provision additional resources to the VM or request additional instances of the application that is seeing decreased performance due to increased demand; of course that one request kicks off a number of processes including checking for capacity, orchestration, configuration, and provisioning).

Private cloud orchestration does not uniquely present this challenge, for any tool that is a cog in the wheel that is IT presents similar challenges. The broader opportunity is, how do we make each wheel turn to propel the vehicle that is our IT organization forward? (Please excuse the analogy.) What many organizations lack is this ability to actually control their infrastructure and the elements therein; in other words, the wheels are there but the drivetrain to make them operate and turn together is the real challenge that is now bestowed upon us.

To see how one customer leveraged VMTurbo in their Orchestration view this excerpt or see how VMTurbo can help accelerate and guarantee the performance of your private cloud deployment.

Leave a Reply

Your email address will not be published. Required fields are marked *