Hi, On Tue, Feb 01, 2022 at 03:38:32PM -0800, Jeff Davis wrote: > On Tue, 2022-02-01 at 20:45 +0800, Julien Rouhaud wrote: > > > > One last thing, did you do some benchmark with a couple custom rmgr > > to see how > > much the O(n) access is showing up in profiles? > > What kind of a test case would be reasonable there? You mean having a > lot of custom rmgrs? > > I was expecting that few people would have more than one custom rmgr > loaded anyway, so a sparse array or hashtable seemed wasteful. If > custom rmgrs become popular we probably need to have a larger ID space > anyway, but it seems like overengineering to do so now.
I agree that having dozen of custom rmgrs doesn't seem likely, but I also have no idea of how much overhead you get by not doing a direct array access. I think it would be informative to benchmark something like simple OLTP write workload on a fast storage (or a ramdisk, or with fsync off...), with the used rmgr being the 1st and the 2nd custom rmgr. Both scenario still seems plausible and shouldn't degenerate on good hardware.
