Great article Jesper, having spent a little time around corporate setting I can't help but thinking that sometimes new decisions are taken with a mixture of the three following reasons:
we've always done it like that
cost related limitations
"Nobody ever got fired for choosing IBM" (replacing IBM with a tech giant of the contemporary era)
None of those is a bad factor to take into account when making a decision but they can creep in complexity without a proper analysis.
The craze around microservices that started a few years ago is an example. Microservices are a good idea, if that is what you need. Otherwise are a very very bad one because they increase complexity of your system n-fold.
Same thing is happening now with Kubernetes. Great idea, until it isn't:
Dan Woods
@danveloper
If I ever recover from this last week of work, I'll tell y'all about how a Kafka outage caused a Kubernetes cluster to self-destruct, resulting in the production of 10s of 1,000s of nodes into a Consul gossip mesh that took down ended up taking out an entire infrastructure.
02:55 AM - 14 Jan 2019
4391598
(and these folks are experts and for them k8s was the right choice)
Great article Jesper, having spent a little time around corporate setting I can't help but thinking that sometimes new decisions are taken with a mixture of the three following reasons:
None of those is a bad factor to take into account when making a decision but they can creep in complexity without a proper analysis.
The craze around microservices that started a few years ago is an example. Microservices are a good idea, if that is what you need. Otherwise are a very very bad one because they increase complexity of your system n-fold.
Same thing is happening now with Kubernetes. Great idea, until it isn't:
(and these folks are experts and for them k8s was the right choice)
We, developers, like shiny things :)
Thanks a lot for your thorough response. I agree totally with your points and especially "Nobody ever got fired for choosing IBM".
Scary tweet with Kubernetes. Some times fancy new solutions are the right decision, but a lot of times they are not.