Picking an IaaS public cloud? Look for 'good bones'

July 12, 2018

A blog post by David Linthicum, managing director, chief cloud strategy officer, Deloitte Consulting LLP

Infrastructure as a service (IaaS) clouds are much the same as houses. They need a good foundation of IaaS services to provide value to your enterprise. For an IaaS cloud, this means it provides both storage and compute, in ways that make them most efficient and useful.

Today, there tends to be a focus on the higher-level platform services, such as machine learning, databases, serverless, etc. In fact, the larger IaaS players now put 5-10 services per week into their cloud services catalog and most of these services have little to do with storage or compute.

These might be ‘want and need’ services, but it’s important to look at the foundation services to ascertain the true quality of a particular public cloud. After all, those are building blocks of all higher-level platform services.

One example of these services is a database. Obviously, a database needs to store data, and thus must leverage the underlying IaaS cloud storage systems, either object, block, or file storage.

No matter how good the database is, if the underlying storage infrastructure does not work well, then performance issues, I/O errors, and other problems will become commonplace. These days the storage systems and database are typically coupled, and a failure of the underlying infrastructure means a failure of the higher-level service. It’s just that simple.

The underlying primitive services need to function well enough to support all higher-level cloud platform services. That’s the bottom-line definition of an IaaS cloud with good bones. It’s something you need to look out for, as well as test, before you migrate existing applications or build net-new applications on a public IaaS cloud.

So, what are some indicators of good bones?

First, understand how the underlying storage and compute systems are architecturally structured. Find out how the public cloud supports tenant management, I/O management, the operating systems that are offered, and the native cloud ops subsystems. Of course, what you look for could be a bit different, considering that each enterprise’s underlying requirements and use of higher-level platform services will vary.

Second, test with sample applications. Move a few application workloads into the public cloud and see how they do in production. Or, build net-new testing programs that simulate what the applications will do in production.

You’ll get quick insight into the IaaS cloud’s ability, or not, to support the application patterns you’ll likely leverage. Also, its ability to support higher-level cloud platform services you’ll likely use in production.

Good bones are not something you hear much about in the world of cloud computing right now. We’re so focused on higher-level services that the underlying IaaS-native services don’t seem to get as much attention. However, ignore the bones and you’ll likely be standing on a shaky foundation.