Re: bstat and estat

"Karsten Farrell" <kfarrell_at_belgariad.com> wrote in message news:rpAA9.111$lZ3.14947696_at_newssvr21.news.prodigy.com...
> Howard J. Rogers wrote:
> > "Karsten Farrell" <kfarrell_at_belgariad.com> wrote in message
> > news:xVzA9.98$FK3.14076239_at_newssvr21.news.prodigy.com...
> >
> >>
> >>Add to the above a couple of other tuning things you can do in your
> >>listener.ora file:
> >>
> >>* If your users connect infrequently, sqlnet might need to spawn a new
> >>dispatcher, which goes away when there's no activity for a while.
> >
> >
> > Excuse me????? Dispatchers are not automatically spawned. The number of
> > dispatchers is governed by MTS_DISPATCHERS (just DISPATCHERS in 9i) and
the
> > number is fixed. The parameter is dynamically adjustable, but not by the
> > Listener, or by "Sqlnet", whatever that means, since sqlnet is just a
> > protocol. And dispatchers don't just "go away" after a while, either.
> >
> > Are you perchance confusing this with the shared server processes, which
are
> > dynamically spawned and destroyed, depending on the load on the system
(and
> > none of which, of course, would make the slightest difference to a
connect
> > time, since connections in MTS are always to a dispatcher)?
> >
> > HJR
> Yes, you are right. I used the wrong term. Instead of "dispatchers", it
> should have been "prespawned dedicated server processes". I was using
> "sqlnet" in the generic sense (maybe not a good idea) of Oracle
> networking (but sqlnet isn't a protocol - it's a piece of the session
> layer of a protocol stack).

Uh huh. It's a proprietary wrapper around the network protocol, sure enough (which is what I think you're saying it is here). Point being, it has no intrinsic smarts, and wouldn't, of itself, know how to spawn daisies or daffodils.

Confusing "dispatchers" with "prespawned dedicated server processes" isn't a use of the wrong term. It's a complete bastardization of the Oracle architecture. It's like saying "I meant to say 'blue' when I wrote 'red'".

> Yes, in a non-MTS environment, connections are handed off to a dedicated
> server process (which is spawned/destroyed). In an MTS environment,
> connections are handed off to a dispatcher, which parcels out the work
> to shared server process.

Which are spawned/destroyed.

Yup, we know all that. Thanks all the same. Feel free to double-check your own knowledge by repeating it out loud.

HJR > Received on Wed Nov 13 2002 - 17:11:12 CST