RE: more good iseries news

Andy,

This is what I was looking for.   There are many things I don't know about
websphere - or processor load sharing for that matter.

I just assumed that any job running would take advantage of any processor
that wasn't busy.

What about the native iseries HTTP server (or even apache)?  Is it able to
share request loads across multiple processors (assuming a single instance
of the server)?

Rick

----original message----
Andy said:

Subject: RE: more good iseries news

My take on the paragraph -

I believe that they had an application server defined within WebSphere.
As they increased the load on the server it took advantage of the second
server well, but received little benefit from subsequent processors.
Perhaps for reason of multithreading or some other circumstance, it had
two 'jobs' which took most of the processors.  Additional processors
would not be of benefit.

At this point, a logical next step is to create another software
instance of that same WebSphere server on the same machine.  For
whatever reason, multiple instances must have created some integration
issues which they wished to avoid, thus they went with loading
completely separate copies of WebSphere on physically separate servers.
WebSphere is designed to expand via multiple instances on the same
hardware, I don't know what the circumstances were here.

That's my guess,
Andy Nolen-Parkhouse



> > >Rick
> > >
> > >The issue of maxing out on capacity was faced early on. Passer said
> > >stress tests were performed at the IBM iSeries headquarters, in
> > >Rochester, Minnesota. From that, it was determined that adding a
> > >second processor would be "an easy fix." The next round of
increasing
> > >capacity is more complicated. "After two CPUs in the iSeries,"
Passer
> > >explained, "the application server has too much overhead and
doesn't
> > >really offer a big advantage over multiple application server
> > >instances. In all cases, it means more CPUs, but there are major
> > >software issues one gets bogged down with."