[rt-users] Performance with many attachments and transactions

[Ben Goodwin]

 > -----Original Message-----
 > From: Jesse Vincent [mailto:jesse at bestpractical.com] 
 > Sent: Tuesday, November 29, 2005 6:38 PM
 > To: Ben Goodwin
 > Cc: rt-users at lists.bestpractical.com
 > Subject: Re: [rt-users] Performance with many attachments 
 > and transactions
 > 
 > On Tue, Nov 29, 2005 at 06:34:40PM -0500, Ben Goodwin wrote:
 > > I hate to chime in with a "me too" especially without hard 
 > data but:

[snip]

Displaying a particular ticket under RT 3.5.6 3 times produces:

%Time ExclSec CumulS #Calls sec/call Csec/c  Name
 14.4   4.858  4.858  51699   0.0001 0.0001  Encode::utf8::encode_xs
 7.79   2.618 11.798 168354   0.0000 0.0001  RT::Record::__Value
 6.40   2.151  3.282 289387   0.0000 0.0000
DBIx::SearchBuilder::Record::_Prim
                                             aryKey
 5.37   1.803  4.388 168354   0.0000 0.0000
DBIx::SearchBuilder::Record::Cacha
                                             ble::__Value
 5.35   1.798  4.655 250916   0.0000 0.0000  DBIx::SearchBuilder::Record::Id
 4.92   1.651  1.651 268566   0.0000 0.0000  RT::Record::_Accessible
 4.79   1.610  2.620 169879   0.0000 0.0000
DBIx::SearchBuilder::Record::__Val
                                             ue
 3.98   1.337 44.079   7556   0.0002 0.0058  HTML::Mason::Commands::__ANON__
 3.36   1.129  1.129 291293   0.0000 0.0000  RT::Record::_PrimaryKeys
 3.24   1.087  3.852 167482   0.0000 0.0000  Encode::decode
 2.97   0.999  1.574 114716   0.0000 0.0000
Cache::Simple::TimedExpiry::fetch
 2.96   0.995  1.608 167481   0.0000 0.0000  Encode::utf8::decode_xs
 2.88   0.966  7.467  51005   0.0000 0.0001  RT::Record::LoadByCols
 2.59   0.869  0.869  51005   0.0000 0.0000
DBIx::SearchBuilder::Record::Cacha
 
ble::_gen_alternate_RecordCache_ke
                                             y
 2.54   0.854  6.253  51005   0.0000 0.0001
DBIx::SearchBuilder::Record::Cacha
                                             ble::LoadByCols

Indeed - the total ExclSec (24.725) is right on target with how long it took
to display the ticket.

I can give some hard details on MySQL but I'm fairly convinced it's not the
database (it's a modified version of my-huge.cnf).

It seems alarming that the function taking the least time is STILL almost a
second long.

System config attached.

	-=| Ben
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: sysconf.txt
URL: <http://lists.bestpractical.com/pipermail/rt-users/attachments/20051207/150a0239/attachment.txt>