Cloud is not all about cost

Hey folks,

I think the DHH post from last year around migrating back to on-prem is a bit misguided.

I don't know if the numbers make sense, but I've looked into them. Maybe I just didn't dig deep enough, but I wonder, was the move to on-prem really cheaper when you had to retrain people to deal with managing servers? Really?

Racking and managing servers is a whole different mindset, and retraining isn’t cheap. That was something I never saw in his numbers. And if they didn’t have to retrain, that seems a bit odd too.

The bigger issue, particularly for most startups and mid-market companies, is that the on-premise approach doesn't make sense because the cloud gives you optionality. You can grow into different services that your cloud provides. Ones that an on-prem data center is never going to provide. You could invest the time to run an equivalent open-source tool on-prem, but that isn’t cheap either.

If you are in the cloud, a managed service which AWS or Google Cloud provides will be much faster to set up and use.

So, if you're iterating on your product at all, don't go on-prem because it’ll be slower to innovate.

It feels like it's that simple.

If you're not iterating on your product, what are you doing?

To get back to the DHH article, maybe they've shoehorned everything into Rails—after all he’s the inventor of that framework. But not everybody can do that.

Other folks have more involved products and more involved infrastructure.

I think that this speed of complex service adoption and delivery is not talked about enough when that article or the idea of moving back to on-prem comes up.

What do you think?

May your product always be iterating,

Matt @ Masterpoint

PS If you want to hear more about my software engineering journey, as opposed to my opinion of DHH’s article on cloud, I was recently on the The Modern Independent podcast.