Hacker News Reader

Comment by stackskipton | Hacker News Reader

Ops person here, VMware is so embedded at these companies that switching away would be like Google saying, no more gRPC, everything is now SOAP. The amount of impacts is just too mind boggling to even consider.

2 companies ago was heavily invested in VMware. It impacted monitoring, backups, deployments, networking, cloud migration and more. I can only shudder at level of effort they might be going through to get off VMware.

Because of that, they probably won’t for years even as Broadcom screws them over.

View on HN

Switching away today is too mind boggling to consider, but switching is purely a technical problem. We can put a price on the costs, and the time. VMWare was founded in 1998 - that means 27 year ago nobody was using it, and in turn we can say took you less than 27 years to get dependent on and - surely you can switch to something else in 27 years. More likely you can switch in 2 years - that is about what it took one company I know of.

Sure but old company cut Ops team to the bone, which is why I don’t work there anymore. So CTO is faced with, pay VMware or cut back on deployments so Ops team can have some breathing room to work on migration to whatever they pick.

... Google did say "No more Oracle EBS" and switched entirely to SAP. It took multiple years, and it was not a small effort, but there was the will, and a way was found.

sigh I didn’t work at FAANG type, we don’t hire FAANG skill level. Whatever Google did is completely irrelevant to this conversation and most other conversations.

In my experience, FAANG are no more skilled than anyone else. The main thing is that they seem to do a better job of not hiring complete duds, so their average cleverness may be higher, but I’ve worked with brilliant people at every regular shop.

I agree but complete duds can really drag your team down as you end up spending a ton of time trying to fix their shortcomings and amount of code written on the Ops side to prevent them from completely screwing everything up is mind boggling.

The bias with technical people is thinking every problem is technical

VMWare may have hiked the prices and might be an important dependency but at a certain point it is cheaper to sue and/or switch from them.

Seems that they have gone way past this point

AI-assisted migration of glue boilerplate code transforms this mind-boggling amount of impact into a two-year project, max.

What code? An on-prem VMWare deployment is all about hardware, storage, networking, and fuck-tons of planning, budgeting, and approvals. There is little to no customer-written code in a typical VMWare farm, except maybe some Ansible or whatever for minor customizations and automation.

Only to the extent that the team is competent enough to properly test all that boilerplate code, which is very far from a given. A relative of mine in IT has had a large internal-tooling migration get dragged on for years by the persistently bug-ridden code of one of the groups working on it.

The AI can’t migrate the knowledge of all the people that have to operate your on-prem VM deployments.

These days it is really hard to figure out if comments like these are real or satire.

Reply on news.ycombinator.com