james-server-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Bernd Fondermann <bf_...@brainlounge.de>
Subject Re: Long Running Test
Date Fri, 17 Feb 2006 23:33:03 GMT
Steve Brewin wrote:
> I'm not sure there is an average usage scenario. At one extreme James is
> deployed by home users and small business where the number of users and
> volumes are low but the mail they send is most likely destined for a remote
> destination. At the other extreme an objective has always been to be able to
> handle the traffic levels that apache.org deals with, a large volume of
> external mail. Another enterprise level scenario is dealing with large
> volumes of local mail.

The short time goal would be to make the objective performance of two 
distinct James builds (like 2.2.0 vs. 2.3.0a1) comparable. This could 
lead to shorter release cylces.

The longer time goal is to max out James to identify needs for 
optimization under extreme conditions. For that, the stress test itself 
must be proven and stable, which of course is not the case as of today.

> 
> For the apache.org scenario it might be possible to have access to the
> current traffic statistics and build a simulation around these. Does anyone
> know how we might obtain these?

That would be cool. Wouldn't infrastructure know that?

> 
> Another approach might be to stress test local and remote delivery
> separately. In each case ramping up the volumes until failure. This would
> reveal what breaks when due to the lack of what resource. Once the failure
> point is identified, we would want to explore the relationship between mail
> volume, size and composition around this point to understand its true
> nature.
> 
> I guess I'm saying that we need to understand the edge of the envelope of
> what James can do so that we can make adjustements to broaden the envelope
> in the future.
> 
> Thanks again.
> 
> Cheers,
> 
> -- Steve
> 
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: server-dev-unsubscribe@james.apache.org
> For additional commands, e-mail: server-dev-help@james.apache.org
> 
> 


---------------------------------------------------------------------
To unsubscribe, e-mail: server-dev-unsubscribe@james.apache.org
For additional commands, e-mail: server-dev-help@james.apache.org


Mime
View raw message