Re: Deleting older versions in unique indexes to avoid page splits

On Sat, Oct 24, 2020 at 2:55 AM Simon Riggs <simon@2ndquadrant.com> wrote:
> The problem I highlighted is that the average UPDATE latency is x2
> what it is on current HEAD. That is not consistent with the reported
> TPS, so it remains an issue and that isn't obvious.

Why do you say that? I reported that the UPDATE latency is less than
half for the benchmark.

There probably are some workloads with worse latency and throughput,
but generally only with high contention/small indexes. I'll try to
fine tune those, but some amount of it is probably inevitable. On
average query latency is quite a lot lower with the patch (where it is
affected at all - the mechanism is only used with non-hot updates).

-- 
Peter Geoghegan