SPOF – How a brilliant developer doesn’t make a good product
One of my many colleagues within the company I’m working for once built a very cool an powerful service some years ago. So powerful and architecturally elegant that most of our websites use it and are heavily dependent on it. I wouldn’t know to point out a single flaw in this technology.
Other than available maintenance ressources.
The service I’m talking about does scale quite a bit and is in a redundant setup. So with single point of failure I do not refer to the service alone. It does stutter sometimes but runs ok in general.
However when something breaks it’s the same genius mind over and over again that fixes it. Just the one. There is no backup. And he moved on within the organisation taking on other responsibilities and so forth.
Now he’s going to stay off work on maternity leave. Who could blame him? But who will replace him in the meantime?
Who will replace him when he’s calling in sick or when he decides to seek employment elsewhere?
A product like this service simply isn’t ready for production only when it’s implemented. It needs to be run, maintained and developed further and for that you need to spread the knowledge to more than one pair of shoulders. If you can not or are not willing to spend some ressources on running a product you should not build it yourself in the first place.