3 ms·
The older I get, the more I've slid towards the opinion that I'd much rather configure cloud SaaS via terraform than waste days of work maintaining a self-hoste
by amyjess 4y ago
The older I get, the more I've slid towards the opinion that I'd much rather configure cloud SaaS via terraform than waste days of work maintaining a self-hosted application.
- Tehchops 4y agoYea reading HN comments gives me the impression most of the readership has never had to manage anything of complexity beyond their "homelab".
- ozim 4y agoWhen 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.
- Tehchops 4y agoYep. 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.
- raffraffraff 4y agoAgree 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.
- ushakov 4y agome too but for me the SaaS offering has to be either: a) based on open-source b) interoperable (S3, MQTT) otherwise prepare to be screwed over by greedy corporates