geronimo-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Maxim Berkultsev" <maxim.berkult...@gmail.com>
Subject A few performance notes for MySQL and Derby
Date Fri, 16 Jun 2006 08:26:12 GMT
Hello all,

I've found a difference in performance while working with MySQL and
Derby databases in Geronimo 'steady state' operational phase.

Sample 'create-find' CMP beans scenario takes a one and a half time
longer for MySQL on create operations while for Derby contrary
performance results appear at find operations.

Here is the data:

1000 CMP EJBs;
JMeter as a workload generator;
single thread load;
'find' expects to extract all 1000 entities;
the 'find' loop is repeated for 1000 times.

The results are:

Create operation with Derby: throughput = 860/60 requests per sec;
average response time=66ms.
Create operation with MySQL: throughput = 1270/60 requests per sec;
average response time=44ms.

Find operation with Derby: throughput = 7820/60 requests per sec;
average response time=6ms.
Find operation with MySQL: throughput = 5200/60 requests per sec;
average response time=9ms.

Thanks for everyone who could enlighten the nature of the difference.
(I suppose it is rooted in TranQL...) Are there any issues in
connector DD to make application with MySQL faster?

The results do not pretend to be 'too' precise, I've just repeated the
trials several times and hope to get them correct.

Regards,
Maxim Berkultsev, Intel Middleware Products Division

Mime
View raw message