Re: Compression of full-page-writes

On Sat, Feb 1, 2014 at 10:22 AM, Bruce Momjian <bruce(at)momjian(dot)us> wrote:
> On Fri, Oct 11, 2013 at 12:30:41PM +0900, Fujii Masao wrote:
>> > Sure. To be honest, when I received the same request from Andres,
>> > I did that benchmark. But unfortunately because of machine trouble,
>> > I could not report it, yet. Will do that again.
>>
>> Here is the benchmark result:
>>
>> * Result
>> [tps]
>> 1317.306391 (full_page_writes = on)
>> 1628.407752 (compress)
>>
>> [the amount of WAL generated during running pgbench]
>> 1319 MB (on)
>> 326 MB (compress)
>>
>> [time required to replay WAL generated during running pgbench]
>> 19s (on)
>> 2013-10-11 12:05:09 JST LOG: redo starts at F/F1000028
>> 2013-10-11 12:05:28 JST LOG: redo done at 10/446B7BF0
>>
>> 12s (on)
>> 2013-10-11 12:06:22 JST LOG: redo starts at F/F1000028
>> 2013-10-11 12:06:34 JST LOG: redo done at 10/446B7BF0
>>
>> 12s (on)
>> 2013-10-11 12:07:19 JST LOG: redo starts at F/F1000028
>> 2013-10-11 12:07:31 JST LOG: redo done at 10/446B7BF0
>>
>> 8s (compress)
>> 2013-10-11 12:17:36 JST LOG: redo starts at 10/50000028
>> 2013-10-11 12:17:44 JST LOG: redo done at 10/655AE478
>>
>> 8s (compress)
>> 2013-10-11 12:18:26 JST LOG: redo starts at 10/50000028
>> 2013-10-11 12:18:34 JST LOG: redo done at 10/655AE478
>>
>> 8s (compress)
>> 2013-10-11 12:19:07 JST LOG: redo starts at 10/50000028
>> 2013-10-11 12:19:15 JST LOG: redo done at 10/655AE478
>
> Fujii, are you still working on this? I sure hope so.

Yes, but it's too late to implement and post new patch in this
development cycle of 9.4dev. I will propose that in next CF.

Regards,

--
Fujii Masao