The Multi-Cloud Myopia

Ahead of the East African regional Cloud and Cybersecurity summit, slated on the 19th March 2020 at the Serena Hotel, Ian Onyango, the Systems Engineer, Digital Infrastructure at Dimension Data talks to CIO East Africa about the concerns around cloud services, in what he terms the multi-cloud myopia.

In theory, the idea of infrastructure that can be deployed between different cloud providers is a wonderful concept. Who wouldn’t want to seamlessly migrate workloads between cloud providers, on impulse? Something similar to right-clicking & doing a VM migration from one host to another? Only this time it would be from one cloud platform to another.

To serve their own interests, third-party software vendors want you to believe a lot of things that are simply not sustainable in practice. Their ability of oversimplify things, baffles even the most naive of us. They would have you believe that a tiger is just a scaled-up house cat with “anger management issues”.

In my experience, executing a multi-cloud strategy absent of a compelling business reason, is quite frankly an ill-advised default. You’ll adopt the tiger, try to pet it, and end up losing a limb or two. Let me explain.

It’s Tremendously Costly

Five years ago, maintaining a layer of cloud agnosticism made sense. It was not at all clear who the major players were going to be, what the economics would look like in the future, and how painful data migrations would become. Over half a decade later, we have answers to a lot of these questions.

Every CIO I’ve met that who insisted on cloud provider agnosticism did so at tremendous cost — either in actual dollars, technical overhead, or operational complexity. The costs of maintaining a multi-cloud environment pales in comparison to the amount of work it takes to actually pull off a provider migration when the need arises — which it rarely does! You would ideally have to train (or hire) an internal team to implement, support and integrate workloads on 2-to-3 different clouds. All because of chasing discounts?

The math of “save 50 cents an hour on compute” falls flat when the statement finishes with “…and spend $2000 moving data to that provider so those slightly cheaper VMs have something to chew on.” So simply moving your workloads to save on costs will not work in the long term. You’ll potentially save 10% on your workload for a guaranteed multiple of the development and maintenance cost. Which gets me to my second point.

Undifferentiated Technical Nightmare

Technically, multi-provider integration is a barefoot walk in the park — after it’s been strewn with glass and set on fire. My head physically hurts every time I have to think of how to migrate the hundreds of virtual machines from one cloud provider to another. The cloud providers don’t natively provide any tools to migrate workloads (vms) out of their environment. And I don’t blame them. It’s just good business. No one wants to give their customers an easy way out of their platform. They want to make their services as sticky as possible.

For general purpose workloads equally distributed across different providers, do tell, how exactly will your team migrate Amazon Machine Instances (AMI) VMs to Azure VHD VMs to Google’s Cloud Engine? I’m not saying it’s not possible, just that it’s not worth the hustle. You’re essentially doing a lot of undifferentiated work. Is that where you want to spend your innovation energy? Perpetually converting workloads back and forth is just about the worst use of expensive engineering time that I can imagine. Don’t start down this path!

The Opportunity Cost

Every provider of note currently offers a VM instance, a load balancer, a managed database, a Kubernetes offering etc. That gets you most of where you need to be, provided that your application conforms to “traditional” architectural models. However, in a multi-cloud environment, your infrastructure will only be as good as the lowest common denominator. To successfully implement this strategy, you’ll have to deploy only the services that have similar features across all the other providers. That means that you are essentially giving up the all the benefits of selecting one cloud platform and sticking to it.

What you’re giving up is the differentiated services that the providers are racing to deploy. It’s widely accepted that GCP are leaders in the AI/ML space, while Azure is probably best suited for entreprises that are strategically committed to Microsoft technology with pre-existing enterprise agreements.

A better strategy is to focus on your needs

Buckle down. Select one provider who meets your needs. Then build fully integrated services that take complete advantage of their differentiating features. They’ve spent billions to develop these services that give them that extra competitive edge. No need for you to reinvent the wheel. (Unless of course, you have serious self-esteem issues and this constant need to pointlessly prove yourself).

Focus on providing the best solutions to your clients. Build solutions that deliver unique value to all your customers and partners. That’s how you win in this digital age where practically every Tom-Dick-&-Oduori with a credit card can become your competition overnight.

So if multi-cloud is crap, (and it is), why are so many people talking about it? The only folks pushing for it are third-party vendors who stand to lose everything if multi-cloud doesn’t take off. At the end of the day, everyone is trying to sell you something. Most of these vendors are pushing this multi-cloud narrative because if you go all-in on one provider, they will have nothing to sell you e.g. that god-awful cloud management platform that promises to present all your cloud services on “a single pane of glass”.

Pick one provider and go all in. “Kama mbaya mbaya”.

Ian Onyango is the Systems Engineer, Digital Infrastructure at Dimension Data.