Here's the thread describing the changes: https://www.postgresql.org/message-id/20200301083601.ews6hz5dduc3w2se%40alap3.anarazel.de …
-
-
Show this thread
-
There's more to come. But this is as much as I could get to pass tests, and not look too embarrassing tonight.
Show this thread
End of conversation
New conversation -
-
-
Here's a more granular graph, that also includes an additional improvement. The latter is not recomputing the snapshot, which now is fairly easy due to the other changes. https://twitter.com/AndresFreundTec/status/1234037875030650881 …pic.twitter.com/0Z9ZcQlBvx
Show this thread -
The dip for master is interesting, but looks unrelated to the connection scalability itself. Within that band the linux scheduler uses more IPIs, which triggers hypercalls, which are expensive. Below/Above that it doesn't do that at the same rate.
Show this thread
End of conversation
New conversation -
-
-
The X axis spacing is, uh, off. I'll fix that another day. With more granularity at lower client counts too.
Show this threadThanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
What is the hardware configuration for this benchmark? Especially, storage device of the
$PGDATA directory. -
It's an azure f72sv2. Two Xeon Platinum 8168. It's a memory resident workload (scale 100), so storage doesn't play a role.
- 1 more reply
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.