The Reality of Modern IT Operations and Pareto
by Jonathan Reeve September 6th, 2013
The keynote on Day Two of VMworld (and, yes, the “hang space” turns out to be a great place to watch the keynotes), VMware talked about the “Reality of Conventional IT Operations.”
In their on-stage demonstration, VMware went on to show “automated remediation” with workflows between vCenter Operations Manager and vCloud Automation Center (vCAC).
At VMTurbo, we agree that the approach to IT Operations needs to change. It starts with a clear understanding of the problem you’re trying to solve. This problem is Intelligent Workload Management, which, simply stated, is assuring application service levels while utilizing the virtual infrastructure as efficiently as possible. Only by solving this problem can you break the cycle VMware shows in their presentation.
Now, the concept of “automated remediation” has been around IT Operations for a long time, but all threshold (even “dynamic threshold”) approaches to triggering automated remediation are flawed in the same key respect: They fail to understand the consequences of their actions. While it may seem like a good idea to spin up a new VM or resize it in isolation, the reality is that, just like ripples on a pond, a change to one part of the virtual infrastructure can have effects on the other interconnected parts. If we don’t understand these effects, how can we be sure that the action being proposed will be ultimately beneficial?
A little bit about Pareto … Pareto was an Italian engineer who understood that actions have consequences, and as a student of economics and game theory, came up with the concept of “Pareto Optimality:” An allocation of resources where it is not possible to make anyone better off without making someone else worse off.
VMTurbo takes the work of Pareto to heart, and through the application of market economics produces a broad set of automated actions, including placement, sizing and capacity decisions (adding or removing capacity). The recommended actions drive the virtual infrastructure to the “desired state,” in which application performance is assured while infrastructure resources are utilized as efficiently as possible.
Furthermore, this goes beyond remediation and into the realm of prevention (wouldn’t it be better to avoid a problem in the first place?). By continually tuning the environment to its desired state or “happy place,” VMTurbo avoids the need for remediation in most cases.
So, when someone talks to you about automated remediation, make sure you ask the golden question – does the solution understand the consequences of that action?