I see a problem in your results.

You have a 41,87% iowait and only 20MB/s read and a marginal write.

4 cases :
 * An untypical select (I have try the sum to avoid temp table
creation and select * from table) which lead to random reads
(btrees?). Can you post your query and maybe table creation please
(maybe the generation script)?
 * A very poor hard drive or misconfigured.
 * The database file which is fragmented.
 * You have another idea

I tried on two computers, using jdk 6 from sun and the openjdk. Same
results.

I try dropping cache (thanks for the tips) with
watch -n 1 "echo 3 > /proc/sys/vm/drop_caches"
I doesn't change things so much.
Nevertheless I was able to test to give enough system cache to load
the whole data file. iowait @ 0%. Doesn't change performances so much
as iowait was low already.


On 9 jan, 15:31, Peter W <[email protected]> wrote:
> Here's my figures when running a typical SELECT (see first post).
> Flushed disk caches before (seehttp://linux-mm.org/Drop_Caches),
> shows I/O bottleneck:
>
> avg-cpu:  %user   %nice %system %iowait  %steal   %idle
>            6,40    0,00    2,46   41,87    0,00   49,26
> Device:            tps    MB_read/s    MB_wrtn/s    MB_read    MB_wrtn
> sdb             227,00        19,64         0,88         19          0
>
> Running the query again (when the 1,5gb table is cached by os) shows
> user cpu bottleneck:
>
> avg-cpu:  %user   %nice %system %iowait  %steal   %idle
>           45,77    0,00   10,95    0,00    0,00   43,28
> Device:            tps    MB_read/s    MB_wrtn/s    MB_read    MB_wrtn
> sdb               0,00         0,00         0,00          0          0
>
> So I'd say, depending on how much memory you have, large tables can
> have big I/O bottlenecks.
>
> What do you think?
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "H2 
Database" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to 
[email protected]
For more options, visit this group at 
http://groups.google.com/group/h2-database?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply via email to