Bacula-users

Re: [Bacula-users] interesting news about MYSQL

2013-09-24 18:10:31
Subject: Re: [Bacula-users] interesting news about MYSQL
From: Phil Stracchino <alaric AT metrocast DOT net>
To: bacula-users AT lists.sourceforge DOT net
Date: Tue, 24 Sep 2013 18:07:11 -0400
On 09/24/13 17:20, compdoc wrote:
> The most obvious improvement in the new release is speed. MySQL engineer
> Tomas Ulin says MySQL 5.7 is 95 per cent faster than MySQL 5.6 and 172 per
> cent faster than version 5.5. The new version can achieve a peak throughput
> of over 500,000 queries per second on the Sysbench point select benchmark,
> and thanks to code contributions by Facebook engineers, it can now process
> new connections to the database as much as 64 per cent faster.

The question that has to be asked, though, is "under what test
conditions".  I know I've followed up some of the improved-speed claims
in previous releases and found out that what they meant was that, to
pull out general-recollection numbers, MySQL 5.5 scales well to 16 cores
then falls over, while MySQL 5.6 scales to 64 cores before it falls
over, and MySQL 5.6 on a 64-core server is [mumble] percent faster than
5.5 on a 16-core server.  But that's not really a fair comparison.  It
would be more fair to compare the two on 16-core servers, because a
customer who's running 5.5 on an 8-core rented server today isn't going
to go out and rent a 128-core server tomorrow to run 5.7.  And when you
do *that*, comparing apples to apples, the difference is typically only
a few percent.

Which stands to reason when you think about it; if we're talking core
for core, within limits that the older version can handle, to "process
new connections to the database as much as 64 per cent faster" means
that logins have to be processed in a third of the time.  Which would be
a pretty earthshaking level of optimization.

My company doesn't have many customers running 5.6 yet, but some of the
ones we do have are complaining that MySQL 5.6 is actually slower, core
for core and RAM for RAM, than 5.1 on older hardware with the same data.


-- 
  Phil Stracchino, CDK#2     DoD#299792458     ICBM: 43.5607, -71.355
  alaric AT caerllewys DOT net   alaric AT metrocast DOT net   phil AT 
co.ordinate DOT org
  Renaissance Man, Unix ronin, Perl hacker, SQL wrangler, Free Stater
                 It's not the years, it's the mileage.

------------------------------------------------------------------------------
October Webinars: Code for Performance
Free Intel webinars can help you accelerate application performance.
Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from 
the latest Intel processors and coprocessors. See abstracts and register >
http://pubads.g.doubleclick.net/gampad/clk?id=60133471&iu=/4140/ostg.clktrk
_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users