Hacker News new | ask | show | jobs
by ozim 1428 days ago
When I was younger I thought it is cool to be admin and have access to stuff, run things on my own being like Neo in Matrix.

Then I had admin access to wi-fi router in dorm, it stopped being funny when people shitty laptops could not connect and they would blame you or for any kind of connection issue for that matter :)

Then I had a job with support duty, 1 A.M. calls because you need to check logs are not cool.

Now I don't want to have any privileges unless I really cannot do my work. If I don't have to run some application and someone else is getting called for outage or I can say that is our provider can't do much - I am definitely making company I work for - pay to another company and stay out of equation.

If I run some software for myself by myself I am not going to pay for that - but as soon as other people are involved I would tell them to go and buy that service. If it is work related even more so.

1 comments

Yep. Again, the nuance of this point is lost on the "homelab" crowd, but if you're in a software business, your software is what your business is.

Not running a monitoring platform, not running a logging system, not building your own PaaS. Outsource this as much as possible to 3rd-party agents with a sizable economy of scale.

Nearly all engineering teams(outside of FAANG) simply aren't resourced to re-invent the wheel over and over, especially when it means providing after-hours support for the wheel as well.

Agree to a point. I'd say that "nearly all" teams outside those in medium to large companies aren't resourced enough to maintain existing well-packaged wheels. It's a gradient, not a binary. If you've got someone who's capable of deploying EKS and setting up a CI/CD that lets developers move quickly, it mightn't be a big deal for them to deploy the Grafana stack (loki, mirmir, grafana) and have logs and metrics on the cheap. Log/metric/APM services can get ludicrously expensive and unless you're in fintech, you might only need them for debugging and capacity planning.