qpid-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "John O'Hara" <john.r.oh...@gmail.com>
Subject Re: [jira] Created: (QPID-10) Broker throughput falls off with transactions
Date Fri, 29 Sep 2006 09:44:46 GMT
How badly does it fall off?

One would expect throughput to be a lot lower if you're doing 1 TX per
message.  About a third the performance.
If you do 1 TX for every 10 messages, performance should go up again.
If you do 1 TX for every 100 messages, performance will be faster for
persistent messages (get that!).

In real systems which care about persistence and guarantees, it is critical
to batch transactions to maintain throughput.
Its also critical to hold off writing to disk till the last possible moment
in the transaction to increase concurrency.

Try it with an RDBMS of your choice - its just a fact of life :-)


On 29/09/06, Martin Ritchie (JIRA) <qpid-dev@incubator.apache.org> wrote:
> Broker throughput falls off with transactions
> ---------------------------------------------
>                  Key: QPID-10
>                  URL: http://issues.apache.org/jira/browse/QPID-10
>              Project: Qpid
>           Issue Type: Bug
>           Components: Java Broker
>             Reporter: Martin Ritchie
>          Assigned To: Martin Ritchie
> When running the IBM PutGet test in transaction mode (-tx) throughput
> falls away.
> Investigation is needed to see why the broker is getting bogged down.
> --
> This message is automatically generated by JIRA.
> -
> If you think it was sent incorrectly contact one of the administrators:
> http://issues.apache.org/jira/secure/Administrators.jspa
> -
> For more information on JIRA, see: http://www.atlassian.com/software/jira

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message