3 ms·
I have lots of experience with Postgres (including scaling somewhat from a web app perspective) and have nothing really bad to say about it. It scaled well. T
by epi0Bauqu 19y ago
I have lots of experience with Postgres (including scaling somewhat from a web app perspective) and have nothing really bad to say about it. It scaled well. The key (at least for me) is to make sure indexes are in memory, which I think will be the case no matter what DB you use.
- e1ven 19y agoHave you used server-clustering with pgsql? Slony doesn't do multi-master, and requires manual failover; I understand the technical reasons for this (The slave may not be up to date, so data will be lost, etc), but we're looking for other options. pg_cluster and it's ilk seem under tested, but I'd be VERY interested in your experiences, if any. Clustering under mysql has always been fairly trivial, so it'd be nice to find a decent solution.
- babul 19y agoI would be interested to hear about this too.
- epi0Bauqu 19y agoI have found that the situations you really need clustering are few and far between in the world of web apps. We had over 20M users and we're running fine on one 2U database server. It had 6 disks (RAID 5, with two hot spares) and 16GB of memory. The indexes all fit in memory. The acquiring company of our business immediately decided they wanted more hardware for clustering, and set up Slony. I was semi-involved in it, and I witnessed no problems. On a bigger scale, I understand that skype uses pgsql and obviously does clustering: https://developer.skype.com/SkypeGarage/DbProjects/SkypePostgresqlWhitepaper https://developer.skype.com/SkypeGarage/DbProjects/SkypePost... Also check out http://highscalability.com/postgresql-high-availability-websites http://highscalability.com/postgresql-high-availability-webs... and http://highscalability.com/skype-plans-postgresql-scale-1-billion-users http://highscalability.com/skype-plans-postgresql-scale-1-bi...