On Sat, Oct 8, 2016 at 11:27 PM, Fabien COELHO <coe...@cri.ensmp.fr> wrote:
> Hello Amit.
> Also, given the heavy UPDATE nature of the pgbench test, a non 100% default
>>> fill factor on some tables would make sense.
>> FWIW, sometime back I have seen that with fill factor 80, at somewhat
>> moderate client counts (32) on 192 - Hyper Threaded m/c, the
>> performance is 20~30% better, but at higher client counts, it was same
>> as 100 fill factor.
> The 20-30% figure is consistent with figures I collected 2 years ago about
> fill factor on HDD, see the beginning run of:
> Although I found that the advantages is reduced after some time because
> once a page has got an update it has some free space which can be taken
> advantage of later on, if the space was not reclaimed by vacuum.
> I cannot understand why there would be no advantage with more clients,
> Alas, performance testing is quite sensitive to many details:-(
The current status of the patch and recent mail thread discussion doesn't
represent the same.
Closed in 2016-11 commitfest with "returned with feedback" status.
Please feel free to update the status once you submit the updated patch.