Re: Tracking of page changes for backup purposes. PTRACK [POC]

Hello Robert,

> I think this doesn't really show much because it's apparently limited
> by the speed of fsync() on your filesystem. You might try running the
> test with synchronous_commit=off.

You are right, synchronous_commit=off revealed a noticeable performance
degradation. Also I realized that using log_statement=all was not very
smart as well. Here are the results.

10.1, ptrack_enable=false, synchronous_commit = off

transaction type: <builtin: TPC-B (sort of)>
scaling factor: 1
query mode: simple
number of clients: 4
number of threads: 4
duration: 300 s
number of transactions actually processed: 1713550
latency average = 0.700 ms
latency stddev = 0.434 ms
tps = 5711.822110 (including connections establishing)
tps = 5712.251807 (excluding connections establishing)

10.1, ptrack_enable=true, synchronous_commit = off

transaction type: <builtin: TPC-B (sort of)>
scaling factor: 1
query mode: simple
number of clients: 4
number of threads: 4
duration: 300 s
number of transactions actually processed: 1691011
latency average = 0.710 ms
latency stddev = 0.380 ms
tps = 5636.691378 (including connections establishing)
tps = 5636.730514 (excluding connections establishing)

10.1, without ptrack, synchronous_commit = off

transaction type: <builtin: TPC-B (sort of)>
scaling factor: 1
query mode: simple
number of clients: 4
number of threads: 4
duration: 300 s
number of transactions actually processed: 1843623
latency average = 0.651 ms
latency stddev = 0.589 ms
tps = 6145.395486 (including connections establishing)
tps = 6145.441431 (excluding connections establishing)

--
Best regards,
Aleksander Alekseev