Earlier quoted context omitted.
Beware, the first link appears to be AI slop with at least some bogus information. For example, it says "While PostgreSQL 15 introduced basic WAL compression", but WAL compression has been around since before 15.
That's a bit unfair, don't you think? https://www.percona.com/blog/new-wal-archive-module-library-...
Memory Size Matters to PostgreSQL
11–13 of 13 posts
Re: Memory Size Matters to PostgreSQL
#12For beginners, it is a huge footgun, that makes people assume bad performance while evaluating. For the experienced PG admin, it is an annoiance and a time waster. Oh, the VM just gained 64GB RAM? PG will sit there and stare at it.
Apart from that, basically everyone starts with the PG guidelines or a generated template(25% for this, 25% divided by number of sessions for that). Then you keep wondering how much performance you left on the table.
Re: Memory Size Matters to PostgreSQL
#13I don't fully understand this article but this point stuck out as probably fractally wrong: Modern DDR4 memories have a theoretical throughput of 25-30 GB/s. This is more realistically ranging between 5-10 GB/s. With a 100 GB full packed shared buffer the time required to perform one single full scan ranges between 3 and 20 seconds. Obviously DDR5 now exists and servers have multiple memory channels giving total memo…
Even saying 25-30GB/s is weird. DDR4-3200 is ~26GB/s per channel, and is the upper end of what you'll see on ECC DDR4. DDR5-5600 is common now, and is ~45GB/s. Zen 2/3 Epycs on SP3 have 8 channels, Zen 4/5 Epycs on the SP5 have 12 channels per socket, and with both you get to have two sockets. That'd be ~410GB/s on dual socket SP3 and ~1080GB/s on dual socket SP5. So, yeah, RAM goes brrr.