37signals understands downtime.


37signals is a smart company. Not only have they seen incredible success with their product line, but they run an extremely successful blog, have gotten more press coverage than most companies 50 times their size, and have a very good reputation. They also read other blogs and knew not to make the same mistake that DreamHost made when they had some issues with their service.

Last week, all of the products/services made by 37signals were offline for about two hours. The problem was with their load balancer affecting their connections. It seems like a pretty technical issue that most customers probably wouldn’t understand if explained fully to them.

37signals did a lot of things right with their announcement:

  • They said what happened.
  • They addressed a big concern early on. Was data lost? No.
  • They apologized.
  • They said they weren’t happy about what happened and apologized again.
  • They offered compensation to affected clients, even though they don’t have a formal SLA plan that requires them to do so. They also made this process easy.
  • They said they had the best service provider out there, but the company dropped the ball (see below).
  • However, 37signals repeatedly said the problem is ultimately their problem and their fault. They claimed full responsibility.

Commenters on the post pointed out that 37signals only used one load balancer (when two are often standard), which is their fault, not their service provider’s fault. An employee from 37signals replied and said they will be adding one. As mentioned above, they claimed full responsibility for the downtime. It doesn’t matter who’s fault it is, but it is their problem.

37signals was also pretty responsive to comments on TechCrunch. They updated their status web site frequently during the outage. Basically, 37signals handled the situation as well as any company I’ve seen.

There are very few companies who do things as well as 37signals did with this situation. From what I’ve read, how 37signals handled this situation is how they handle most others – very well. Kudos to 37signals.

2 Responses to “37signals understands downtime.”

  1. Nic said:

    Jan 24, 08 at 12:49 pm

    It was absurdly negligent for 37 Signals, who is constantly heralded as the paradigm of hosted on-demand aps, to have a single point of failure. Their outage was a huge embarrassment and then they crucified their hosting provider.
    The point is, unless the entire power grid goes offline their service shouldn’t have been effected.

  2. Service Untitled said:

    Jan 24, 08 at 11:52 pm


    That’s probably right. It was their fault (as they admitted) for having a single point of failure. It’s hard to tell exactly what their hosting provider did wrong based on the limited details, but it probably would not have happened (or have been a big deal) if they had multiple point of failures.