At Adaptive Computing, policy-based optimization is what we do. As in any subject area, it’s important to separate the myths from the facts. Today, I’ll discuss a few myths that relate to the important topic of policy-based optimization.
Myth #1 – I get policy-based optimization from my virtualization vendor.
This myth is, perhaps, the most common of the three myths I will discuss. While virtualization vendors provide a lot of great tools, policy-based optimization isn’t one of them. Things like VMWare’s DRS are hard-coded to perform certain behaviors, and can’t be adjusted to use different policies or metrics. True policy-based optimization means that for each different set of computing resources, I can utilize different policies as appropriate. It also means that I can feed my custom metrics into the policies, so that the optimization is suited to my environment, not some generalized model. Lastly, the optimization must happen automatically, not through manual migrations or similar processes.
Myth #2 – I can manually implement policy-based optimization.
Sometimes, people think that they can get the benefits of policy-based optimization through old-fashioned hard work. Manual practices like periodically reviewing the environment and making necessary adjustments just don’t work very well in the long term.
Consider for a moment the benefit that you get from having a thermostat attached to your furnace. You set a desired temperature on the thermostat, and it turns the furnace on and off as needed, even when you are asleep. Imagine how cumbersome it would be to manually turn the furnace on when it got too cool and turn it off again when it was too hot. Your full-time job would be babysitting the furnace.
Policy-based optimization works like a very intelligent thermostat. You decide the policies you want to apply to your compute environment and the optimizer continuously adjusts things to enforce your policies. Even when you are sleeping, your environment is kept in an optimal state. Trying to do this manually makes about as much sense as having a furnace without a thermostat.
Myth #3 – I don’t really need policy-based optimization.
This myth may actually be true, in a few cases. If you only have a dozen machines in your computing environment, you can probably manage them manually. However, as your environment grows to hundreds or thousands of machines, manual optimization becomes impossible. The cost and time involved are simply too great.
Similarly, if you don’t care about wasting money, you also might not need policy-based optimization. Our customers save millions of dollars by using policy-based optimization. While wasting millions may not matter to a small handful of organizations, there is tremendous pressure in the general IT world today to reduce costs and deliver more. By utilizing hardware much more efficiently, policy-based optimization allows organizations to purchase much less hardware, saving significant amounts of money. Coupled with the time savings from automating manual processes, the benefits of policy-based optimization are too great to ignore.