On Tue, May 7, 2013 at 9:23 PM, Peter Geoghegan <pg@heroku.com> wrote:
On Tue, May 7, 2013 at 12:48 PM, Jeff Janes <jeff.janes@gmail.com> wrote: > Anyway, I don't see this behavior change when turning on wal_debug and > looking in the logfiles for 'xlog flush request' messages.
That could have everything to do with the hardware you're using. In general, the higher the cost of an fsync, the more useful it is to amortize that cost among concurrently committing transactions.
True, but that is going to happen with a proportional increase in throughput, which he reported not seeing. At least originally. I'm not sure what to think now.