Presenter: Brian L Wong
Contrary to often repeated “wisdom,” capacity planning in a cloud is not obsolete. In fact, it is more relevant than ever. But when it is done, how it is done, the data upon which it is done, by whom it is done and the impact of doing it, are all different in a cloud than in traditional data center environments. In a cloud, more capacity planning is done after the launch of an application than before. The impact of making a mistake is, usually, lower and the number of environments to be sized is generally greater than historically seen in data centers. The data used to inform capacity decisions is quite different. Even the form of the applications is different enough to influence the capacity planning process. Containers, fleets of microservices, function-as-a-service, fully managed services and many other new concepts call for new sources of data and more analysis by both human capacity planners and their ML assistants.
This talk outlines the next-generation computing environment from an observability, capacity and analytical perspective, and speculates on the form and value of capacity planning in these environments.
About the Presenter
Brian Wong is a Technology Fellow at Capital One. A systems engineer by trade, a common theme running through his career is an interest in performance, capacity and benchmarking, whether of CPUs, systems, storage or applications. He is responsible for cost and effectiveness of cloud operations in Capital One’s infrastructure. Previously a Distinguished Engineer at Sun Microsystems, he has worked in servers, operating systems and storage. While not obsessing over the differences between throughput, latency and efficiency in an application, or between microseconds, milliseconds and nanoseconds in a benchmark, he can be found observing one or more of the following: the benchmarking of Formula One or Indy cars (ie races), large metal hulky things from the past (such as steam locomotives or ocean liners) or Corgis, often with a camera in hand.