Jay,

Thank you for the updated instructions, I got the benchmark to run with
minimal pain. I got the following results:

mysql> SELECT r.version, c.name, i.concurrency, AVG(i.tps) as TPS FROM
sysbench_config c  NATURAL JOIN sysbench_runs r NATURAL JOIN
sysbench_run_iterations i WHERE r.server = 'drizzled' AND r.version =
'trunk-1039' GROUP BY r.version, c.name, i.concurrency;
+------------+------------------------+-------------+------------+
| version    | name                   | concurrency | TPS        |
+------------+------------------------+-------------+------------+
| trunk-1039 | innodb_1000K_readonly  |           2 | 456.845000 |
| trunk-1039 | innodb_1000K_readonly  |           4 | 679.435000 |
| trunk-1039 | innodb_1000K_readonly  |           8 | 730.975000 |
| trunk-1039 | innodb_1000K_readonly  |          16 | 747.365000 |
| trunk-1039 | innodb_1000K_readonly  |          32 | 722.605000 |
| trunk-1039 | innodb_1000K_readonly  |          64 | 696.020000 |
| trunk-1039 | innodb_1000K_readonly  |         128 | 683.105000 |
| trunk-1039 | innodb_1000K_readonly  |         256 | 661.630000 |
| trunk-1039 | innodb_1000K_readonly  |         512 | 586.905000 |
| trunk-1039 | innodb_1000K_readwrite |           2 | 305.745000 |
| trunk-1039 | innodb_1000K_readwrite |           4 | 513.070000 |
| trunk-1039 | innodb_1000K_readwrite |           8 | 601.260000 |
| trunk-1039 | innodb_1000K_readwrite |          16 | 577.165000 |
| trunk-1039 | innodb_1000K_readwrite |          32 | 564.715000 |
| trunk-1039 | innodb_1000K_readwrite |          64 | 554.000000 |
| trunk-1039 | innodb_1000K_readwrite |         128 | 528.050000 |
| trunk-1039 | innodb_1000K_readwrite |         256 | 488.005000 |
| trunk-1039 | innodb_1000K_readwrite |         512 | 451.735000 |
+------------+------------------------+-------------+------------+

This was on:

x86_64 opensuse 11.1
single cpu (Intel(R) Core(TM)2 Quad CPU    Q6600  @ 2.40GHz)
4GB RAM

I couldnt run larger then a concurrency of 1024 I got the following error
which looks like a ulimit problem I need fix on my machine, although max
open files I upped to 20K for the test.

2009-05-27 16:37:29,647 INFO: Running sysbench config innodb_1000K_readonly
for concurrency at 2048 - iteration 0.
/home/jdaly/drizzle/repos/trunk-sysbench-r1000/drizzled/drizzled: Forcing
close of thread 1983  user: ''
Traceback (most recent call last):
  File "/usr/lib64/python2.6/logging/__init__.py", line 754, in emit
    msg = self.format(record)
  File "/usr/lib64/python2.6/logging/__init__.py", line 637, in format
    return fmt.format(record)
  File "/usr/lib64/python2.6/logging/__init__.py", line 425, in format
    record.message = record.getMessage()
  File "/usr/lib64/python2.6/logging/__init__.py", line 295, in getMessage
    msg = msg % self.args
TypeError: %d format: a number is required, not tuple


It may be worthwhile to run the benchmarks on a Solaris host it may expose a
call somewhere that is platform inefficient in a new library.

--Joe Daly



On Fri, May 15, 2009 at 8:47 PM, Jay Pipes <[email protected]> wrote:

> Joe Daly wrote:
>
>> Hi Jay,
>>
>> Im trying to get up to speed, on the development here so excuse my silly
>> question.
>>
>
> Hi! Sorry for th delayed reply...
>
>  What were these benchmarks against?
>>
>
> These benchmarks are against each BZR revision of Drizzle on one of our
> benchmarking machines (16-core x86 box).
>
> You can read more about our benchmarks on this thread:
>
> https://lists.launchpad.net/drizzle-discuss/msg03687.html
>
> Cheers!
>
> Jay
>
>  thanks
>> --Joe Daly
>>
>> On Wed, May 13, 2009 at 9:21 AM, Jay Pipes <[email protected]> wrote:
>>
>>  Hi all!
>>>
>>> Sorry for the delay in getting the data for analysis of our benchmark
>>> revision history!  Attached is a tarball containing a single SQL script
>>> to
>>> populate a database with our data for revisions 980 through 1008.
>>>
>>> To setup, do:
>>>
>>> $> tar -xzf drizzle_stats.tar.gz
>>> $> mysqladmin --user=root --password=$yourpass create drizzle_stats
>>> $> mysql --user=root --password drizzle_stats < drizzle_stats.sql
>>> <enter your password>
>>>
>>> Here are the sysbench tables:
>>>
>>> mysql> desc sysbench_runs;
>>> +-----------+-------------+------+-----+---------+----------------+
>>> | Field     | Type        | Null | Key | Default | Extra          |
>>> +-----------+-------------+------+-----+---------+----------------+
>>> | run_id    | int(11)     | NO   | PRI | NULL    | auto_increment |
>>> | config_id | int(11)     | NO   |     | NULL    |                |
>>> | server    | varchar(20) | NO   |     | NULL    |                |
>>> | version   | varchar(50) | YES  |     | NULL    |                |
>>> | run_date  | datetime    | NO   |     | NULL    |                |
>>> +-----------+-------------+------+-----+---------+----------------+
>>> 5 rows in set (0.00 sec)
>>>
>>> mysql> desc sysbench_config;
>>> +-----------+--------------+------+-----+---------+----------------+
>>> | Field     | Type         | Null | Key | Default | Extra          |
>>> +-----------+--------------+------+-----+---------+----------------+
>>> | config_id | int(11)      | NO   | PRI | NULL    | auto_increment |
>>> | name      | varchar(255) | NO   |     | NULL    |                |
>>> +-----------+--------------+------+-----+---------+----------------+
>>> 2 rows in set (0.01 sec)
>>>
>>> mysql> desc sysbench_run_iterations;
>>> +---------------------------+---------------+------+-----+---------+
>>> | Field                     | Type          | Null | Key | Default |
>>> +---------------------------+---------------+------+-----+---------+
>>> | run_id                    | int(11)       | NO   | PRI | NULL    |
>>> | concurrency               | int(11)       | NO   | PRI | NULL    |
>>> | iteration                 | int(11)       | NO   | PRI | NULL    |
>>> | tps                       | decimal(13,2) | NO   |     | NULL    |
>>> | read_write_req_per_second | decimal(13,2) | NO   |     | NULL    |
>>> | deadlocks_per_second      | decimal(5,2)  | NO   |     | NULL    |
>>> | min_req_latency_ms        | decimal(10,2) | NO   |     | NULL    |
>>> | avg_req_latency_ms        | decimal(10,2) | NO   |     | NULL    |
>>> | max_req_latency_ms        | decimal(10,2) | NO   |     | NULL    |
>>> | 95p_req_latency_ms        | decimal(10,2) | NO   |     | NULL    |
>>> +---------------------------+---------------+------+-----+---------+
>>> 10 rows in set (0.00 sec)
>>>
>>> To see a view of the data which I've been showing on this list, I use
>>> this:
>>>
>>> mysql> SELECT r.version, c.name, i.concurrency, AVG(i.tps) as TPS
>>>   -> FROM sysbench_config c
>>>   -> NATURAL JOIN sysbench_runs r
>>>   -> NATURAL JOIN sysbench_run_iterations i
>>>   -> WHERE r.server = 'drizzled'
>>>   -> GROUP BY r.version, c.name, i.concurrency;
>>>
>>> which shows you data in this format:
>>>
>>> *************************** 1. row ***************************
>>>   version: fix-bitset-regression-1009
>>>      name: innodb_1000K_readonly
>>> concurrency: 2
>>>       TPS: 1080.706667
>>>
>>> Have fun. :)
>>>
>>> Cheers,
>>>
>>> Jay
>>>
>>> _______________________________________________
>>> Mailing list: 
>>> https://launchpad.net/~drizzle-discuss<https://launchpad.net/%7Edrizzle-discuss>
>>> <https://launchpad.net/%7Edrizzle-discuss>
>>> Post to     : [email protected]
>>> Unsubscribe : 
>>> https://launchpad.net/~drizzle-discuss<https://launchpad.net/%7Edrizzle-discuss>
>>> <https://launchpad.net/%7Edrizzle-discuss>
>>> More help   : https://help.launchpad.net/ListHelp
>>>
>>>
>>>
>>
>
> _______________________________________________
> Mailing list: 
> https://launchpad.net/~drizzle-discuss<https://launchpad.net/%7Edrizzle-discuss>
> Post to     : [email protected]
> Unsubscribe : 
> https://launchpad.net/~drizzle-discuss<https://launchpad.net/%7Edrizzle-discuss>
> More help   : https://help.launchpad.net/ListHelp
>
_______________________________________________
Mailing list: https://launchpad.net/~drizzle-discuss
Post to     : [email protected]
Unsubscribe : https://launchpad.net/~drizzle-discuss
More help   : https://help.launchpad.net/ListHelp

Reply via email to