A foolproof way to understand cloud optimization

We’ve discussed the notion of cloud architecture optimization here in the past. Now it’s time to understand how it’s measured. You really have no way to prove your architecture isn’t optimized unless you do an audit, which includes a review of the solution’s approach and any attached costs.

In the past, the people who built and deployed cloud solutions were reluctant to have their choices questioned. These days, because we want the most value from the cloud solutions, many have changed their minds about questions and oversight—or more often, company leadership changed their minds for them. Many of the projects I take on these days focus on review and improvement audits rather than on build and migrate deployments.

Once everything works in cloud architectures, you can deploy and operationalize. Just because it works does not mean that it’s optimized. If you look at the differences between your architecture and one that’s optimized, you could have a “working” solution that costs you millions of dollars each week.

To present this visually, see the figure below. Note that positions 1 and 37 are the least optimized. They cost more money and are the most inefficient.

When we look at each side, note that cloud solutions can be under-utilized or overutilized, such as with containers and serverless computing. Those on the left of the chart might not have included enough containers, whereas those scoring on the right have used containers too much. The point of optimization is to be in position 19, where we use the right number of containers to make the most of costs and solution efficiency.

optimal cloud architecture IDG

Of course, you can leverage this metric for any holistic architecture or all configured technology. You could even apply it to microarchitectures, such as a few applications moving to serverless or containers, for instance. Note that including polynomial views of the data for smoothing creates a curve that’s more likely to reflect typical real-life behavior. In the real world, this data never actually runs in a straight line.

Copyright © 2020 IDG Communications, Inc.

Source link