On 10/23/2013 07:51 PM, Peter Geoghegan wrote:
> On Wed, Oct 23, 2013 at 4:46 PM, Josh Berkus <josh@agliodbs.com> wrote:
>> So you're suggesting that instead of storing the aggregates as we
>> currently do, we store a buffer of the last N queries (in normal form)
>> and their stats? And then aggregate when the user asks for it?
> No, I'm not. I'm suggesting storing the query texts externally, in a
> file. They usually use 1024 bytes of shared memory per entry,
> regardless of how long the query text is. This would allow
> pg_stat_statements to store arbitrarily large query texts, while also
> giving us breathing room if we have ambitions around expanding what
> pg_stat_statements can (optionally) track.
>
> Having said that, I am still pretty sensitive to bloating pg_stat_statements.
>
Me too. I think min, max and stddev will have a fairly small impact, and
give considerable bang for the buck. Not so sure about the other
suggestions. And of course, memory impact is only half the story - CPU
cycles spent is the other part.
I'll be quite happy if we can get around the query text length limit. I
have greatly increased the buffer size at quite a few clients, in one
case where they run some pretty large auto-generated queries and have
memory to burn, up to 40k.
cheers
andrew
