Hacker News Reader

Comment by vegetablepotpie | Hacker News Reader

This is how things worked 15 years ago, before the “cloud” hype took off. In my opinion, things worked great. You could get stuff done.

I’m intrigued that there’s a “movement”. I suppose that’s what you need to get news articles and promote change. What’s old is new.

View on HN

I do mostly remember and encounter with on-premise datacenters the following pains

- Small selection of services (databases, queue's,...) If a service was not yet in house, brace yourself for a year of meetings, enterprise contracts, security audits,...

- Overprovisioning of everything as it was a weekly to monthly process to acquire new resources

- lots of half-assed documentation with many manual steps being missed. Sometimes the DEV environment was so different from the PROD environment you couldn't reliable trust deployments.

- lots of outdated software and servers with tooling that promised to ease all of that while delivering none of it.

- different teams for different components, all with their own procedures, zero API's and you hoping that the one guy that actually helps you is not on a holiday.

- Very strange limitations because how all the different vendors had their own set of restrictions and zero incentive to adopt another vendors solution.

It's interesting how scarcity of work experience in a technology can drive up advertised job opportunities which in turn drives technology adoption as everyone seeks to boost their resume with the in demand skills.

Things did not worked great, you should see the bad practices about ci/cd, security etc ... let the dev ssh root into a linux box, 1000 days uptime, no patch...

Exactly the same things now happen in your cloud provider. Now you just don't see them.

You might see them. But it's far easier for a dev team to take control of the ops side in the cloud with IaC, API's and automated tooling like Terraform. It's still very hard to find companies that do on-prem, API first.

You didn't get the point. Your cloud provider can do things that you don't see and can't control. If you think all cloud provider's employees follows best security practices because they told you so, then good luck.

I briefly worked for a medium sized web company (millions of registered users, 100's of employees, pretty large infrastructure.) It was hosted on AWS. I would regularly SSH into production instances that had 1000 days uptime, no patches, OS releases that from 5+ years ago, etc.

Oh yeah, they were still running Python 2 for parts of the main application. It was literally like traveling back a decade. I found dependencies that hadn't been updated in 10 years.

Reply on news.ycombinator.com