3 ms·

(Not a DB Engineer but poke at enough different DBs to armchair semi-confidently) In the case of Postgres I'd expect to see it indeed be mostly about random up

(Not a DB Engineer but poke at enough different DBs to armchair semi-confidently) In the case of Postgres I'd expect to see it indeed be mostly about random updates to leaves. I suppose Partitioning could come into play but I'm not sure how you'd set it up in a way that was beneficial and dynamic enough for the situation in the first place. But, It's worth remembering that most other DBs (Oracle, MSSQL, MySql/etc) do have clustered indexes, and in some cases you can't just flip over to Postgres. (Even if you want to!)

SigmundA 4y ago
both MSSQL and Postgresql support fill factor as well, which nobody seems to mention. Setting a lower fill factor leaves more room in the pages to avoid page splits with random writes. Also random writes can be beneficial in that you are not hot spotting the last page as you are with sequential ID's. Its not black and white, there are some good advantages with random ID's but you need to manage them properly.