I’m always concerned about how enterprises pick their cloud technology configuration. It’s easy to pick virtual platforms from a menu of hundreds of configurations that include memory, storage, CPUs, I/O, and networking. Whatever you pick will probably work, but it won’t be cost-optimized.
Enter the new cloud-selection systems. A few new entries to the market promise to remove the guesswork when it’s time to pick a cloud configuration that’s right for specific use cases, industries, and application types. They also tout the ability to analyze existing on-premises systems and determine the proper equivalents in the cloud.
Should you push decisions about configuration and cloud selections to these IT decision-making systems?
The answer is that it really depends on the data—both the data the decision-making systems have and the data you have. For example, if you do your own proper analysis of your existing on-premises systems, or perhaps your systems that are already in the cloud, you’re likely to find the exact CPU configuration needed, storage, memory, and networking, yourself.
That’s because you provided the decision-making system with the correct data in the right level of detail t make it simple for it to create an equivalent configuration. By the way, you could get the same results with some basic math, which I have done many times.
However, where these tools do provide value is in the selection of brands beyond configurations. While the data from your existing systems can easily lead to a configuration on a single cloud, which cloud should you choose? Is it Microsoft, Amazon Web Services, Google, or something else? Moreover, each cloud sells resources differently, and at different costs.
As a result, you may get different optimizations proposed by these tools based on a combination of different cloud services and their best technology approaches, such as containers or serverless.
The critical factor is that these systems must have the right data. I remember metrics systems of the past that were consistently wrong because the data they used was crap. The same could occur here.
So, if you’re going to use these decision-making systems, put in some test data where you know the correct answer and see what it says. While you’re likely to have an exact match for an answer, if you get a different result , it should at the very least make sense.
Other factors to consider include the possibilities that you may not want to or can’t use some cloud providers or can’t use more than one, such as due to skill limitations or other internal barriers to building a best-of-breed system.
I suspect that decision-making systems will become more popular as enterprises look to take the guesswork out of cloud platforms and configuration selections. They will also help with the optimization of your cloud platform, which will be an ongoing task. Just remember: Garbage in, garbage out.
