Post #3741460

2026-07-11 10:28 UTC

Sometimes newer is not better. At least not for the things that matter to you. I benchmarked two OVH VPS in the same Zurich location. My current 2025 model with a 500 Mbit network against the newer 2027 model with 1 Gbit. Same AMD EPYC Genoa cores, same provider, nearly identical sysbench CPU scores. Then the surprises started. The older box pushes 6,612 MiB/s in memory tests, the newer one 932 MiB/s. That is 7x. Random 4K reads: 1.32 million IOPS against 832k. And the old model is even about 1 EUR cheaper. For a Mastodon instance, databases, and Docker workloads, memory and disk beat extra bandwidth every time. So the shiny new VPS gets demoted to reverse proxy and backup duty, and this instance stays exactly where it is. Benchmark before you migrate. #SelfHosting #Mastodon #Fediverse #VPS #OVH #Benchmarks #Homelab #Blog #MastoAdmin

Replies (2)

@gryzor@androiddev.social 2026-07-11 13:00

@dima@dol.social could it be a dual channel memory issue?

Open ##3741459
@dima@dol.social 2026-07-11 11:56

The numbers behind the previous post, for the curious. Same fio and sysbench setup on both boxes, Debian 13, direct=1, libaio. Random read 4K, the workload databases actually live on: 1.41M IOPS on the 2025 box against 822k on the 2027 one. RAM bandwidth: 64,038 MiB/s against 11,906. That is 5.4x. Single thread CPU: 4,504 against 4,147 events/sec. Same EPYC Genoa on paper. And here is my favorite part. The old box ran all of this while carrying my entire production: Mastodon, Postgres, Valkey, a pile of Docker containers, 17 days of uptime. The new box was empty and idle. The loaded veteran still won almost everything. #SelfHosting #VPS #Benchmarks #fio #sysbench #Homelab #MastoAdmin #Fediverse

Open ##4262042