The VMware Distributed Resource Scheduler handles typical workload spikes with aplomb to keep your virtual environment running smoothly. But what if VMware DRS were part Nostradamus, and had the foresight to prepare for regularly occurring disruptions?
Mike Preston, systems analyst at the Hastings and Prince Edward District School Board in Ontario, thought DRS could be enhanced with that foresight to recognize problematic patterns and act accordingly before virtual resources showed signs of stress. He had noticed patterns in his environment -- for example, at regular intervals DRS would take action and, at others, Distributed Power Management (DPM) would power on more hosts to handle increasing workloads.
Around that time, VMware was soliciting submissions to its 2012 Open Innovation Contest, asking users to suggest ways to improve the company's virtualization portfolio; VMware would then fashion the winning idea into a fling.
Preston, co-leader of the Toronto VMware User Group, submitted his "Proactive DRS/DPM w/ vCenter Operations" idea. He suggested that statistics VMware vCenter Operations (vCOPs) gathered could be used to build "some sort of appliance [or] script that can hook into both the vCenter Operations APIs as well as the vSphere APIs and merge these two technologies together."
How did the idea for Proactive DRS come about?
Mike Preston: I've been a [VMware vCenter Operations] user ever since the very first beta, after VMware acquired Integrian, so I've had plenty of time to see it mature and pick out spots where it could be further integrated into the core vSphere products. For the most part I watched vCOPs learn what is 'normal' in our environment, and I also watched how DRS reacted to that.
One particular workload requires very heavy CPU usage every morning, and every morning I watched how DRS would react to this event. I just thought, since vCOPs already knew that this was going to happen, why couldn't we inform DRS of this ahead of time and be a bit more proactive in resource management?
Did you think your idea would win the fling contest?
Preston: Not in the least! There were a lot of very well thought-out and innovative entries dealing with everything from backups to snapshot management.
Were you involved with VMware's engineers in the development of Proactive DRS?
Preston: I did converse a little bit back and forth through email, answering questions, providing comments, etc. Nothing very technical in nature, mostly high-level information.
What are some other flings that have caught your eye?
Preston: Hands down, if I had to choose one fling out of them all, it would by Onyx. It has saved me so much time when I couldn't figure out how to do something in PowerCLI or vCenter Orchestrator. Aside from that, I love the VMware OS Optimization Tool for its functionality, and I also am very intrigued by the vBenchmark tool and the data it returns.
Are you planning to submit an idea for the next fling contest?
Preston: I am definitely planning on submitting another idea for this year's contest. I'll keep my idea secret for now!