Might be me getting old but just copying what “big 4” are using seems to be the trend even on smaller projects. Running Kubernetes or mesos to manage a tiny cluster, running hadoop, spark … infrastructure to analyse 10 gig data set, using casandra, hbase etc to store it. WTF?
2 Likes
Nah 
All you can do is advise people to the best of your ability and if they choose to ignore that advice then let them carry on 
But how small are you talking? Or they always going to be small projects or do they have the potential to grow/become very busy? If so I don’t see any harm in being prudent if it’s possible and without being too much of a pain.
2 Likes
Yes, was just talking to a friend about this. Happens a lot in web development. People think that because Facebook needs GraphQL/Relay their app does too, even if’s just a CRUD app.
This may be too cynical, but I read this on HN today:
Many programmers have no real substance in their work – the job is trivial – so they have too much time on their hands, which they use to dwell on “API design” and thus monstrosities are born
Keeping things simple is the hardest thing about programming.
2 Likes
One of my colleagues used React to serve a static page 
6 Likes
They have potential to run out of money overspending on useless tech 
I mean using kubernetes on 4 node cluster, deploying hadoop to handle data set that you can analyse in R on a desktop. You do not really need kubernetes or mesos to scale homogenous workload even at very large scale. The potential for screw ups grows exponentially with complexity increase.
When doing a small project, one might use a large or new tech stack to learn joe everything might fit together. But yes, you really do overcomplicate it for yourself, which both results in a high technical debt and often a high monetary debt/upkeep as well.
It usually is a lot easier, when the need arises, to make a simple system more complicated, than making a complicated system more simple.
KISS(keep it simple, stupid!) and YAGNI(you aren’t gonna need it) apply here.
1 Like
Yes. I just finished my Phoenix app so I was learning about how to deploy it. During my research, I was reading about releases, docker, kubernetes, continuous integration and so on. My brain stopped working due to information overload. Until now I haven’t deployed my app yet.
1 Like
I just use exrm (should update to the new one by the same author…), it works very well.
1 Like
I skipped exrm and jumped to trying distillery. Aren’t you using edeliver?
edeliver does not support Windows, and I must support both Linux and Windows at work, so it is pure exrm for me (which supports everything fantastically, edeliver is… limited).
