On 31.01.2012 01:35, Simon Riggs wrote:
> The plan here is to allow WAL flush and clog updates to occur
> concurrently. Which allows the clog contention and update time to be
> completely hidden behind the wait for the WAL flush. That is only
> possible if we have the WALwriter involved since we need two processes
> to be actively involved.
> The theory behind this is clear, but needs some explanation.
> There are 5 actions that need to occur at commit
> 1) insert WAL record
> 2) optionally flush WAL record
> 3) mark the clog AND set LSN from (1) if we skipped (2)
> 4) optionally wait for sync rep
> 5) remove the proc from the procarray
> Notice that step (2) and step (3) are actually independent of each other.
> So an improved design for commit is to
> 2) request flush up to LSN, but don't wait
> 3) mark the clog and set LSN
> 4) wait for LSN once, either for walwriter or walsender to release us
That seems like a pretty marginal gain. If you're bound by the speed of
fsyncs, this will reduce the latency by the time it takes to mark the
clog, which is tiny in comparison to all the other stuff that needs to
happen, like, flushing the WAL. And that's ignoring any additional
overhead caused by the signaling between processes. If you're bound by
CPU capacity, this doesn't help at all because it just moves the work
Anyway, this is quite different from the original goal and patch for
group commit, so can we please leave this for 9.3, and move on with the
review of pending 9.2 patches.
In response to
pgsql-hackers by date
|Next:||From: Simon Riggs||Date: 2012-01-31 08:02:24|
|Subject: Re: Group commit, revised|
|Previous:||From: Marko Kreen||Date: 2012-01-31 07:19:22|
|Subject: Re: Speed dblink using alternate libpq tuple storage|