Basecamp decided to leave the cloud after spending $3.2 million in one year

[Alfonso Maruccia]

In a nutshell: Basecamp announced it was leaving the cloud in October 2022. The company CTO has opened up and detailed how much they were spending on renting computing resources in just a single year, and it's way too much money. Especially compared to what the company can achieve by using its own servers.

Basecamp said it was leaving cloud services because "renting computers is (mostly) a bad deal for medium-sized companies like ours." Cloud companies were making "obscene margins" on computing resources they were renting out, and the savings promised in reduced server complexity never materialized.

David Heinemeier Hansson, CTO of 37signals/Basecamp and creator of the server-side application framework Ruby on Rails, has detailed how much money the company spent on cloud services in 2022. Overall, the web-based project management tool squandered $3,201,564 in just one year -- which means $266,797 of fixed costs per month.

The biggest cost for Basecamp's cloud provisioning was Amazon AWS S3, the remote storage platform where the company spent $907,838 to archive over 8 petabytes of files. Hansson notes the setup was using a "dual-region replication strategy," so that 37signals' biggest product would be resilient against an entire AWS region disappearing.

Basecamp also spent $759,983 (or $63,331 per month) on AWS EC2 and EKS computing services, which Hansson compared to the $1,287/mo cost needed to get the same computing resources on "insanely powerful" physical servers you can purchase from Dell.

Even bringing the cost of cloud resources down to $3.2 million took a lot of work, Hansson revealed.

"The ops team runs a vigilant cost-inspection program, with monthly reporting and tracking," the Basecamp CTO said, and they also entered into a "long-term agreement on committed usage and reserved instances as part of a private pricing agreement." A highly optimized budget which could still be "dramatically" cut by moving most of the cloud services to bare metal hardware.

Basecamp will (mostly) abandon the cloud during 2023, but that doesn't mean the company will need to operate its own data center. 37signals will work with a managed hosting company named Deft, leasing "rackspace, bandwidth, power, and white glove service." Which isn't cheap either, Hansson said, but it still is "far, far less than what we spend on the cloud."

Nobody from 37signals will need to visit any data center: "we order from Dell, have it delivered straight to the data center, and then we see the servers appear online, and then we get to work," Hansson said.

Permalink to story.

https://www.techspot.com/news/97300-leaving-cloud-basecamp-spent-32-million-year-rent.html

 

[wiyosaya]

Surprise, surprise! Jumping on a fad costs more than buying the equipment to do the same for your own company.

 

[bviktor]

s/Cloud/Amazon/

These aren't equivalent.

All that happened was they chose the wrong vendor. I hate Amazon exactly because of it, every kind of action is billed separately. Oh, you want SQL? Cool, we'll bill you on the instance, on storage, on transaction count, on user count, on authentication actions, on network bandwidth, ....... most of which are pretty much IMPOSSIBLE to estimate.

 

[Dimitriid]

Surprise, surprise! Jumping on a fad costs more than buying the equipment to do the same for your own company.

The problem (And mind you, even though I'm quoting you, this is more directed at the article and the logic at hand, not really you specifically) Is that if you just buy a few servers and being generous we assume you already had a ready-to-go server room, it's just going to sit there uselessly if you don't have anyone who can operate it.

And now we get into the real (And yes, still way overpriced but not as much as you'd think) cost of running your own servers which is well, having a professional who can run them.

And let me tell you, just having one guy handle all of the server infrastructure, updates, patches, back up and redundancy strategies, cybersecurity, user access control is not going to be a 1 or 2 persons job: We're talking an entire department. One that's dedicated just to ready up and maintain your hardware to a point where is actually usable and that can come even close to the uptime promises of AWS or Azure.

If you want to talk numbers, that's basically most if not all of your 750k per year on those employees and their benefits, covering for their vacations and downtime and the costs of recruiting more of them once they inevitably move on to some other jobs, retire, get promoted either inside or outside the organization, etc.

I know there's plenty of smaller cloud providers that can get you like 90% of the way there in terms of what AWS or Azure offer and save you a healthy chunk of money and those are probably more cost-effective for a small-to-medium company than moving back to on site. The problem is that AWS and Azure and Google are so big that well, a lot of people working on cloud solutions will probably gravitate towards those solutions and now, you've got to think about your entire organization both present and future and knowing if you want to commit to one of the smaller guys knowing that even though it will save you a lot of money upfront today, it probably will become an issue when it's time to replace all of the employees you have working on your custom/small cloud solution because well the market forces determine that most IT professionals will have lots of prior work experience working with the big cloud providers.

Honestly it sounds flashy to say "Yeah get out of the cloud!" but it has a lot more moving pieces than just "This is how much AWS charge and this is how much buying physical hardware from Dell costs!" that's quite a terrible argument against the cloud imho.

 

[dangh]

Xaas is made to drain every penny from users pockets. Good those guys had the will to fix it

 

[passwordistaco]

The problem (And mind you, even though I'm quoting you, this is more directed at the article and the logic at hand, not really you specifically) Is that if you just buy a few servers and being generous we assume you already had a ready-to-go server room, it's just going to sit there uselessly if you don't have anyone who can operate it.

And now we get into the real (And yes, still way overpriced but not as much as you'd think) cost of running your own servers which is well, having a professional who can run them.

And let me tell you, just having one guy handle all of the server infrastructure, updates, patches, back up and redundancy strategies, cybersecurity, user access control is not going to be a 1 or 2 persons job: We're talking an entire department. One that's dedicated just to ready up and maintain your hardware to a point where is actually usable and that can come even close to the uptime promises of AWS or Azure.

If you want to talk numbers, that's basically most if not all of your 750k per year on those employees and their benefits, covering for their vacations and downtime and the costs of recruiting more of them once they inevitably move on to some other jobs, retire, get promoted either inside or outside the organization, etc.

I know there's plenty of smaller cloud providers that can get you like 90% of the way there in terms of what AWS or Azure offer and save you a healthy chunk of money and those are probably more cost-effective for a small-to-medium company than moving back to on site. The problem is that AWS and Azure and Google are so big that well, a lot of people working on cloud solutions will probably gravitate towards those solutions and now, you've got to think about your entire organization both present and future and knowing if you want to commit to one of the smaller guys knowing that even though it will save you a lot of money upfront today, it probably will become an issue when it's time to replace all of the employees you have working on your custom/small cloud solution because well the market forces determine that most IT professionals will have lots of prior work experience working with the big cloud providers.

Honestly it sounds flashy to say "Yeah get out of the cloud!" but it has a lot more moving pieces than just "This is how much AWS charge and this is how much buying physical hardware from Dell costs!" that's quite a terrible argument against the cloud imho.

This company found a middle ground by owning the servers and paying somebody else for administration.

 

[BadThad]

When they came out with the marketing name "cloud" I almost puked, stupid name and marketing guru's pumped it up as the newest, greatest thing ever! It's client-server - nothing new except now companies are paying big bucks for hosting their server(s).

What happens to your business when some fool accidently cuts your fiber line? That happened three different times at my last company that had it's own data center in-house. We barely knew anything happened and operated right though the outage. My current company uses all hosted servers and "Cloud" services, when the fiber connection/internet go out, it shuts down the entire operation...not even email works, forget about the hosted enterprise systems, they are in "the cloud".

 

[VBKing]

I must admit that when I first saw this headline I was thinking along the lines of Mt. Everest was giving up on having internet connectivity at basecamp and the summit (in the clouds) because it was costing them too much per year. Guess I was way off base.

 

[Burty117]

I recently moved a client from AWS to Azure. Their monthly costs went from £10k a month to £4k. AWS seems to be very expensive.

I wonder if they'd have better luck in Microsoft's cloud.

 

[sorten]

It's very difficult to evaluate their claims based on so few details. It sounds like they did the due diligence and have smart people on their team, and so I'm inclined to believe this was a good decision for them.

There are many ways to save money on your cloud resources by evaluating how those resources are being used. Some companies spin up an EC2 per web service, when they should be running in containers, for example. 8 petabytes (!!) of data is going to cost a fair bit to store and I wonder if they're going to maintain the redundancy that was part of the cost of their AWS setup.

As a developer I love the cloud, but you do need to be careful with costs.