Re: computing completion tag is expensive for pgbench -S -M prepared

On 7 June 2018 at 19:20, Andres Freund <andres(at)anarazel(dot)de> wrote:
> On 2018-06-07 11:40:48 +0100, Simon Riggs wrote:
>> On 7 June 2018 at 11:29, Pavel Stehule <pavel(dot)stehule(at)gmail(dot)com> wrote:
>>
>> >> Do we actually need the completion tag at all? In most cases??
>> >
>> >
>> > affected rows is taken from this value on protocol level
>>
>> I didn't mean we should remove the number of rows. Many things rely on that.
>
> How do you mean it then? We can't really easily change how we return
> that value on the protocol level, and the command tag is basically just
> returned as a string in the protocol. If we were to design the protocol
> today I'm sure we just would declare the rowcount and affected rowcounts
> separate fields or something, but ...

I meant remove the pointless noise word at the start of the tag that
few clients care about.

I was thinking of just returning "SQL" instead, but that wasn't after
much thought.

But now I think about it more returning any fixed string, "SQL" or
"SELECT", in the protocol seems like a waste of bandwidth and a
potential route in to decrypt the stream.

If we're going to compress the protocol, it seems sensible to remove
extraneous information first.

--
Simon Riggs http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services