Postgres Pro
Downloads
Home   >   mailing lists

Re: [JDBC] 42.0.0.jre7, driver logging takes 30-40% of my server'stime - Mailing list pgsql-jdbc

From Daniel Migowski
Subject Re: [JDBC] 42.0.0.jre7, driver logging takes 30-40% of my server'stime
Date
Msg-id 41ED3F5450C90F4D8381BC4D8DF6BBDCAF6E5CFE@EXCHANGESERVER.ikoffice.de
Whole thread Raw
In response to Re: [JDBC] 42.0.0.jre7, driver logging takes 30-40% of my server's time  (Vladimir Sitnikov <sitnikov.vladimir@gmail.com>)
List pgsql-jdbc
Tree view

Hallo Vladimir,

 

OK, I will provide. Just give me a few days, because I am very full of work at the moment.

 

Regards,

Daniel Migowski

 

Von: Vladimir Sitnikov [mailto:sitnikov.vladimir@gmail.com]
Gesendet: Sonntag, 16. April 2017 12:59
An: Daniel Migowski <dmigowski@ikoffice.de>
Cc: Jorge Solórzano <jorsol@gmail.com>; List <pgsql-jdbc@postgresql.org>; Dave Cramer <pg@fastcrypt.com>; Konstantinos Kougios <kostas.kougios@googlemail.com>
Betreff: Re: [JDBC] 42.0.0.jre7, driver logging takes 30-40% of my server's time

 

Daniel,

 

Would you please share your benchmarks?

I don't se how did you arrive at 30% overhead for the logging.

My benchmarks show no overhead attributed to java.util.logging.

 

I did try org.postgresql.benchmark.statement.ProcessResultSet.bindExecuteFetch, JDK 1.8.0_102,

columnIndexType = INDEX, getter = BEST, ncols = 5, nrows = 100, reuseStatement = false, type = BOOL, unique = false.

 

The results are:

 

9.4.1212:

150±8 us / op, 42032 bytes allocated / op

where op is "bind, execute, and fetch resultset of 100 rows, 5 bool columns each"

 

42.0.0:

120±8 us / op, 18032 bytes allocated / op

 

Note: bools are faster in 42.0.0 thanks to Jorge's improvement  in https://github.com/pgjdbc/pgjdbc/commit/e69e4a1d5502319bc810e0e4529611ba52ea386c 

 

String results are exactly the same:

9.4.1212: 

140±6 us / op, 54103 bytes allocated / op

 

42.0.0: 

140±3 us / op, 54103 bytes allocated / op

 

That basically means, logging is a no-op provided it is not activated.

 

JFR recordings show no evidence of java.util.logging on the hot methods list.

logging.isLoggable is rather simple method (it does not walk logger hierarchy), so current logging is expected to be lightweight. 

 

Vladimir

pgsql-jdbc by date:

Previous
From: Vladimir Gordiychuk
Date:
Subject: Re: [JDBC] Logical Replication: adjacent COMMIT messages with thewrong StartLSN
Next
From: Joe Kutner
Date:
Subject: [JDBC] [pgjdbc/pgjdbc] 74ed62: Only resolve hostname if not using a SOCKSproxy