List:Cluster« Previous MessageNext Message »
From:Mikael Ronström Date:August 16 2004 11:59am
Subject:Re: Running sql-bench
View as plain text  
2004-08-15 kl. 02.12 skrev Joseph E. Sacco, Ph.D.:

> Questions:
>
> * Has anyone out there successfully run any/all of the tests found in
> the sql-bench directory in a clustered environment?
>
> * If so, how did it go???
>
> ======================================================================= 
> =
> I have been running some of the benchmarks using the 12Aug04 code.
>
> To avoid editing the test files, a default storage engine type was
> specified when starting mysqld for a cluster:
>
> 	mysqld_safe --default-storage-engine=NDB
>
>
> Some of the tests run in a two computer, 2 DB node cluster, albeit
> slowly.  Others do not.
>
> For example:
>
>         % ./test-wisconsin --socket=/usr/local/mysql/var/mysql.sock
>         Testing server 'MySQL 4.1.4 beta/' at 2004-08-14 19:33:07
>
>         Wisconsin benchmark test
>
>         Time for create_table (3): 58 wallclock secs ( 0.00 usr  0.01
>         sys +  0.00 cusr  0.00 csys =  0.01 CPU)
>

Create table is an operation known to be slow on MySQL Cluster. It has  
never
been designed with performance in the front seat but more in safe  
operation
ensuring that all nodes agree on the time of creation of the table.  
Thus creating
a table is likely to take a few seconds in lead time (not processing  
time).

>         Inserting data
>         Time to insert (31000): 159 wallclock secs ( 2.50 usr  1.14 sys
>         +  0.00 cusr  0.00 csys =  3.64 CPU)
>         Got temporary error 1217 '1217' from ndbcluster at
>         ./test-wisconsin line 179.
>

The problem 1217 is a variant of the problem with too few operations.
Set MaxNoOfConcurrentOperations to a higher value. This part

Rgrds Mikael
PS: I have no experience in this particular benchmark so I don't know  
what is to
be expected. I will also forward the email to our benchmarking experts.

>
> When run against a non-clustered environment:
>
>         % ./test-wisconsin --socket=/usr/local/mysql/var/mysql.sock
>         Testing server 'MySQL 4.1.4 beta/' at 2004-08-14 20:05:12
>
>         Wisconsin benchmark test
>
>         Time for create_table (3):  1 wallclock secs ( 0.00 usr  0.00
>         sys +  0.00 cusr  0.00 csys =  0.00 CPU)
>
>         Inserting data
>         Time to insert (31000): 11 wallclock secs ( 1.23 usr  0.86 sys
>         +  0.00 cusr  0.00 csys =  2.09 CPU)
>         Time to delete_big (1):  1 wallclock secs ( 0.00 usr  0.00 sys
>         +  0.00 cusr  0.00 csys =  0.00 CPU)
>
>         Running the actual benchmark
>         Time for wisc_benchmark (114):  6 wallclock secs ( 3.57 usr
>         0.58 sys +  0.00 cusr  0.00 csys =
>         4.15 CPU)
>
>         Total time: 19 wallclock secs ( 4.82 usr  1.44 sys +  0.00 cusr
>         0.00 csys =  6.26 CPU)
>
>
> Thoughts???
>
> -Joseph
>
> -- 
> Joseph E. Sacco, Ph.D. <jsacco[at]earthlink[dot]net>
>
>
> -- 
> MySQL Cluster Mailing List
> For list archives: http://lists.mysql.com/cluster
> To unsubscribe:     
> http://lists.mysql.com/cluster?unsub=1
>
>
Mikael Ronström, Senior Software Architect
MySQL AB, www.mysql.com

Clustering:
http://www.infoworld.com/article/04/04/14/HNmysqlcluster_1.html

http://www.eweek.com/article2/0,1759,1567546,00.asp


Thread
Running sql-benchPh.D. Joseph E. Sacco15 Aug
  • Re: Running sql-benchMagnus Svensson16 Aug
  • Re: Running sql-benchMagnus Svensson16 Aug
  • Re: Running sql-benchMikael Ronström16 Aug