Re: [REVIEW] Re: Compression of full-page-writes

Hello,

>Updated patches addressing all those things are attached.

Below are some performance numbers using latest patch
20141219_fpw_compression_v9 . Compression looks promising with reduced
impact on CPU usage, tps and runtime.

pgbench command : pgbench -c 16 -j 16 -r -t 250000 -M prepared

To ensure that data is not highly compressible, empty filler columns were
altered using

alter table pgbench_accounts alter column filler type text using
gen_random_uuid()::text

checkpoint_segments = 1024
checkpoint_timeout = 5min
fsync = on

Compression On
Off

WAL generated 24558983188(~24.56 GB)
35931217248 (~ 35.93GB)

Runtime 5987.0 s
5825.0 s

TPS tps = 668.05
tps = 686.69

Latency average 23.935 ms
23.211 ms

Latency stddev 80.619 ms
80.141 ms

CPU usage(user) 916.64s
614.76s

CPU usage(system) 54.96s
64.14s

IO(average writes to disk) 10.43 MB .
12.5 MB

IO(total writes to disk) 64268.94 MB
72920 MB

Reduction in WAL is around 32 %. Reduction in total IO writes to disk is
around 12%. Impact on runtime , tps and latency is very less.
CPU usage when compression is on is increased by 49% which is lesser as
compared to earlier measurements.

Server specifications:
Processors:Intel® Xeon ® Processor E5-2650 (2 GHz, 8C/16T, 20 MB) * 2 nos
RAM: 32GB
Disk : HDD 450GB 10K Hot Plug 2.5-inch SAS HDD * 8 nos
1 x 450 GB SAS HDD, 2.5-inch, 6Gb/s, 10,000 rpm

Thank you,
Rahila Syed