Why we're leaving the cloud

The cloud is a huge convenience and benefit for many. For small applications, renting a cloud shared server is insanely cheap compared to hosting your own. This argument also extends to SAAS services (databases, email, etc). Does it make sense to host my own database, or have someone else do it? For small workloads, it often is much cheaper to put everything (database, webserver, application, etc) on a single cloud server. There is no one size fits all – it’s great to have so many options!


The irony of Basecamp/37signals getting off the cloud is that they offer cloud services to others as their business model :slight_smile:

I agree with you that at various sizes, workloads, and scale, the cloud can be very cost effective. But then at other sizes, workloads, and scale it becomes very not cost effective. Overall I’m very glad we have the public cloud offerings in the market today, it’s another choice available to people and I think that’s good.

1 Like

Some more articles in this series:


It seems that David realizes this to some degree:

I dream of being able to both buy and sell finished business software products.

Then next article in this Saga:

Again, the cloud brings tremendous value – especially to users who are small or have a highly dynamic load. But, if your load is fairly constant, and you have decent scale, it may pay to run the numbers …

Well said!

A podcast version of this:

A number of points are discussed. In summary:

  1. for mid-sized companies with stable loads, running your own infrastructure can save considerable amounts
  2. having a few large companies run most of the Internet runs counter to the original decentralized vision of the Internet
  3. we get conditioned to the mainstream thinking – it’s much “safer” in the hosted cloud, but you still need to know what you are doing, and there are plenty of companies that suffer security problems in the cloud
  4. it is hard to get support from large cloud companies versus a small organization that is more matched to your size.

All of these points resonate with me. Again, the cloud is a great resource for some situations, but it is not the best for every situation. I have also learned that phone support can be critically important – for instance if you get locked out of your account and for some reason email is not working – does not happen very often, but when it does … For this reason, I still use Linode for small-scale critical servers.

This is awesome story and I have read the posts all along the transition

The original X article:


“Optimized our usage of cloud service providers and began doing much more on-prem. This shift has reduced our monthly cloud costs by 60%. Among the changes we made was a shift of all media/blob artifacts out of the cloud, which reduced our overall cloud data storage size by 60%, and separately, we succeeded in reducing cloud data processing costs by 75%.”

But programmers are attracted to complexity like moths to a flame.

You don’t need the cloud to get good uptimes. You need mature technologies run on redundant hardware with good backups. Same as it ever was.

1 Like

Tools like Docker have made it trivial to create closed and isolated systems that can be easily updated and kept secure. Gone are the days of manually tinkering with a box, trying to harden it down. Now all that work has been distributed, and most people run the same handful of base images that have been hardened by a million eyeballs looking in the same place for the same trouble. This is a golden age of secure, baseline computing. We should be celebrating!

A discussion on this topic between Kelsey Hightower and DHH:


Starts around 3:50m.

This is my first experience with Twitter spaces – not super impressed with the listener experience. Good old podcast technology is so much better and is based on decentralized technologies like HTTP and RSS.