tomcat-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Chon-Chon Tang <zt...@weber.lcs.mit.edu>
Subject RE: Tomcat benchmarks from orionserver.com, Please comment. Impor tant
Date Wed, 19 Jul 2000 20:30:31 GMT

On Wed, 19 Jul 2000, Tice, Ed wrote:

> I was not implying that load balancing was solely a Tomcat strength.  My
> point was that with load-balancing available, performance becomes less of an
> issue.  As long as latency is acceptable, throughput can be achieved through
> additional hardware.  I find this preferable to trying to performance-tune
> software.  Would rather the software code be clean and easy to maintain and
> therefore less likely to be buggy.  I will use hardware to increase
> performance.  Yes, I was a theoretical computer science major.
> 
> Ed 

While theoretically true, latency also *appears* to be superior on
Resin/Orion.  

In addition, using hardware to increase throughput is certainly an
option, but you're wasting other resources with each process that might
represent "true" bottlenecks later on.  

For example, a single process pooling ten database connections should
have superior performance than ten processes, each pooling a single
database connection.  Same probably goes for other shared
resources... like shared memory, NFS handles... 


Mime
View raw message