Interesting. The more frequent comment in the comments of the article is “if you use swap, you’re doing it wrong”. What is the current status on performance engineering? Is swap commonly disabled? It does help failing faster, may lower latency, but is it actually disabled in production systems?
Seems like only one person said that, and most people disagree with that opinion. But the argument against swap seems to be specifically a server thing, or for people who are just tinkering with Linux and trying to optimize their performance.