Hi Michael
> > > If I were switch the default, for me zstd goes first, lz4 is a close > > second, and pglz is the last of its class. The only reason why we > > have not chosen zstd for TOAST is the fact that we don't support it > > (trickier to add support for it as we have to preserve on-disk > > compatibility for inline compressed TOAST entries). > > > In short, lz4 is available in many environments, but I'm also ready to > > bet that zstd is equally available in these environments. > Thanks for the feedback. > > Yes, I did consider zstd as well. I agree that zstd can provide a better > compression ratio than lz4, and if we only look at compression ratio, zstd > is > a very attractive choice. > > The reason I started with lz4 was mainly to keep this change aligned with > the > recent TOAST compression change. Since lz4 has already been accepted there > as > the preferred default when available, I thought using the same choice for > wal_compression=on would make the proposal smaller, more consistent, and > easier > to evaluate. so > > Second, WAL compression happens on the WAL insertion path for full-page > images, > so compression speed and CPU overhead matter a lot. lz4 is generally a > good fit > for that trade-off: it still reduces WAL volume, while keeping compression > latency and CPU cost lower than zstd in many workloads. > > That said, I understand the argument for preferring zstd when it is > available. > I have prepared an alternative version where wal_compression=on selects > zstd > first, then lz4 if zstd is not available, and finally pglz as the fallback > Thanks
v002-wal-compression-on-zstd-lz4.patch
Description: Binary data
