2 ms·

Cloud Dependencies Need to Stop F-Ing Us When They Go Down

• 8organicbits 3y ago
  > Immediately, someone suspects it’s a specific third party that you have a hard reliance on. You check the status page of this service, and it says everything is fine. You have to keep searching. Theres a solution for this. Implement a continuous testing suite for your dependencies. I once had an internal service dependency which had poor reliability and was slow to notice or acknowledge problems. Burned too many times, we implemented our own test suite and ran it continuously against their service. Now when they had an outage we could instantly isolate it from our stack. This became our own status page for their service, and they eventually improved their reliability.