Recently saw this blog article from the CTO of Basecamp that is starting to make the rounds in the cloud computing world and thought it would be worth bringing up. It has some points that are often discussed around cloud computing but have a basic misunderstanding of how it works.

The article in question Why we’re leaving the cloud – by David Heinemeier Hansson brings up some pointed ideas around why the cloud is not a good concept for them. I want to break down why at a base level it seems to miss the point of cloud compute.

To get to the point, and I highly suggest reading the article first, I’ll run though his points and give a quick rebuttal around each.

The cloud only excels at tiny workloads and irregular workloads.

He outlines very specifically in his article that it’s great to have growth and that’s the goal, but that cloud doesn’t fit a model that has a steady state. This is very directly combatted with reserved instance and hybrid use concepts. Once you have hit your steady state workload buying the licensing upfront and reserving the machines/services you are using (at least in Azure) will almost always come out cheaper than buying the same hardware on prem. If not just because of the economies of scale the large cloud organizations have compared to a single, smaller buyer.

The cloud doesn’t lead to lower operating/personnel costs.

I don’t know exactly how he came to this conclusion to be honest. As a whole, while you are still using people to manage your cloud services, the end goal is hopefully they are focusing more about how to make it better than how to keep it alive like you would have to do on prem. A huge difference I see in this is now everyone can be an expert on running the service from a higher level than needing to have a separate expert in UPS’s, HVAC, an electrician, and ISPs.

The cloud is just renting computers from someone else.

When highly simplified, sure, when you pay for an IaaS device, you are doing just that, renting a computer. But really you are renting a data center, a power grid, a network connection, storage redundancy, network redundancy, a secure location, ease of creation/ordering, access to our purchase power of hardware, AND a computer. That doesn’t even get into platform as a service offering that adds management features and best practices built in on top of everything else listed.

The cloud is too centralized.

As a concept, I can agree, at this point there are only really two and a half players that genuinely make a mark in the cloud computing space. The answer to this though is doing multi-cloud buildouts and including other trusted data center partners in the mix. The example specifically provided by the article though gets more into a region going down taking out half the internet. That sounds more like an application architecture problem instead of a centralization problem, but that’s just me.

All of this comes around to an important point of the cloud as a whole should be and will be more scalable, resilient, and less expensive than anything a single organization can do on premises at any amount of scale. From the perspective of this article, feel free to use the points I made here but also make sure to see it for what it is: a hit piece without any understanding of the cloud in it’s current state.

Previous Post Next Post