Re: refactoring relation extension and BufferAlloc(), faster COPY

Hi,

On 2023-05-03 18:25:51 +0000, Muhammad Malik wrote:
> Could you please share repro steps for running these benchmarks? I am doing performance testing in this area and want
touse the same benchmarks.
 

The email should contain all the necessary information. What are you missing?


c=16;psql -c 'DROP TABLE IF EXISTS copytest_0; CREATE TABLE copytest_0(data
text not null);' && time /srv/dev/build/m-opt/src/bin/pgbench/pgbench -n -P1
-c$c -j$c -t$((1024/$c)) -f ~/tmp/copy.sql && psql -c 'TRUNCATE copytest_0'


> I've done a fair bit of benchmarking of this patchset. For COPY it comes out
> ahead everywhere. It's possible that there's a very small regression for
> extremly IO miss heavy workloads, more below.
>
>
> server "base" configuration:
>
> max_wal_size=150GB
> shared_buffers=24GB
> huge_pages=on
> autovacuum=0
> backend_flush_after=2MB
> max_connections=5000
> wal_buffers=128MB
> wal_segment_size=1GB
>
> benchmark: pgbench running COPY into a single table. pgbench -t is set
> according to the client count, so that the same amount of data is inserted.
> This is done oth using small files ([1], ringbuffer not effective, no dirty
> data to write out within the benchmark window) and a bit larger files ([2],
> lots of data to write out due to ringbuffer).

I use a script like:

c=16;psql -c 'DROP TABLE IF EXISTS copytest_0; CREATE TABLE copytest_0(data text not null);' && time
/srv/dev/build/m-opt/src/bin/pgbench/pgbench-n -P1 -c$c -j$c -t$((1024/$c)) -f ~/tmp/copy.sql && psql -c 'TRUNCATE
copytest_0'


> [1] COPY (SELECT repeat(random()::text, 5) FROM generate_series(1, 100000)) TO '/tmp/copytest_data_text.copy' WITH
(FORMATtest);
 
> [2] COPY (SELECT repeat(random()::text, 5) FROM generate_series(1, 6*100000)) TO '/tmp/copytest_data_text.copy' WITH
(FORMATtext);
 


Greetings,

Andres Freund