maria-discuss team mailing list archive

[Reindl Harald <h.reindl@xxxxxxxxxxxxx>]

you did not understand the answer!
"ha_innodb.so" is part of *MariaDB 5.5*

InnoDB instead of XtraDB does not make sense in context of MySQL 5.6

>> You can use the Oracle InnoDB instead of XtraDB on MariaDB, it could be interresting to test if it makes a real
>> difference on your specific usecase.
>>
>> To do so, you must put that in our my.cnf file :
>>     ignore-builtin-innodb
>>     plugin-load=ha_innodb.so
>>     innodb

Am 03.07.2013 11:40, schrieb Puff:
> thanks for the fast reply.
> 
> We tried to build MySQL 5.6 and get the InnoDB plugin from this build. It seems the InnoDB is now always statically
> compiled into MySQL and we were not able to get it.
> 
> Related to this topic we found this in the mysql options:  
> 
>   In MySQL 5.6, |*InnoDB*| is the default storage engine and |*InnoDB Plugin*| is not used, so this option has no
> effect. As of MySQL 5.6.5, it is ignored.
> 
>   see http://dev.mysql.com/doc/refman/5.6/en/innodb-parameters.html
> 
> I think using InnoDB from MySQL 5.5 makes no sense since MySQL 5.5 has the same slow results for us.
> 
> *From:*Maria-discuss [mailto:maria-discuss-bounces+werner.puff=gtech.com@xxxxxxxxxxxxxxxxxxx] *On Behalf Of *Jean
> Weisbuch
> *Sent:* Dienstag, 02. Juli 2013 17:45
> *To:* maria-discuss@xxxxxxxxxxxxxxxxxxx
> *Subject:* Re: [Maria-discuss] 5.5 versus 10.0/5.6 performance issues
> 
> Hello,
> 
> You can use the Oracle InnoDB instead of XtraDB on MariaDB, it could be interresting to test if it makes a real
> difference on your specific usecase.
> 
> To do so, you must put that in our my.cnf file :
>     ignore-builtin-innodb
>     plugin-load=ha_innodb.so
>     innodb
> 
> Regards.
> 
> Le 02/07/2013 16:56, Puff, Werner a écrit :
> 
>     Because of performance issues with our DBMS we recently decided to migrate our software to MySQL or one of its
>     forks. We would prefer MariaDB.
> 
>     We created a performance test suite so that we are sure to make the right decision.
> 
>     Beside other values our test measured the average workflow processing time including multiple database
>     transactions. The unexpected outcome was, that we had a big difference in the processing times between 5.5 and
>     5.6 based databases.
> 
>      
> 
>                       average-time [ms]
> 
>                       (lower = better)
> 
>     MySQL 5.6              233
> 
>     Percona 5.6pre         208
> 
>     MariaDB 10alpha        194
> 
>     MariaDB 5.5           1248
> 
>     MySQL 5.5              993
> 
>     (Firebird 2.5         9694)
> 
>      
> 
>     We are bound to very special hardware, the biggest bottleneck there is USB-stick based storage (cannot be
>     changed for now).
> 
>      
> 
>     Our  interpretation of the results and the research we were doing so far is, that the changes within
>     InnoDB/XtraDB storage engine in connection with our "very special" storage solution causes this differences.
> 
>     Can anybody confirm that the changes had a high impact on storage mediums with slow access rates?
> 
>     Or does anybody have an idea if it possible to optimize MariaDB 5.5 (we would like to go with MariaDB) to get
>     the same performance with MySQL 5.6?

Attachment: signature.asc
Description: OpenPGP digital signature