Date: Thu, 27 Jun 2002 21:53:44 -0700 (PDT) From: Matthew Dillon <dillon@apollo.backplane.com> To: nerd@xyz.com Cc: "Gary Thorpe" <gat7634@hotmail.com>, arch@FreeBSD.ORG Subject: Re: Larry McVoy's slides on cache coherent clusters Message-ID: <200206280453.g5S4rioB002878@apollo.backplane.com> References: <200206280035.g5S0ZJmP098253@www.xyz.com>
next in thread | previous in thread | raw e-mail | index | archive | help
:
:In another email on the same thread, Matt Dillon wrote:
:
:>NUMA then becomes just another, faster transport mechanism. That is
:>the direction I believe the BSDs will take... transparent clustering
:>with NUMA transport, network transport, or a hybrid of both.
:
:Matt: If you don't have a single memory immage you don't have NUMA.
:If you do have it then the transport mechanism will be saturated just
:moving "RAM" around and will not be available for network, I/O or
:whatever else.
:
:-michael
:
:michael at michael dot galassi dot org
Well, I wasn't trying to intimate that programs could migrate willy
nilly across the cluster. I was just saying that NUMA is roughly
equivalent to clustering in terms of deterministic program and OS design
(basically what you said about never migrating outside of a quad),
so its better to design an OS or an application to run well in a clustered
environment, which covers NUMA, rather then just a NUMA environment.
My opinion in regards to clustering closely matches what you said.
I am heavily into peer-to-peer quorum-write based infrastructure designs
that can survive machines going on and off line willy nilly in a
cluster without glitching the 'application' running on top of
the infrastructure. I still intend to get the database technology
I developed at Backplane Inc (which is exactly the above) into the
open-source world.
In anycase, I have great respect for Sequent. When I was at Berkeley
working on Postgres we had a (I think) 16 processor (486 based) sequent
and it blew away everything else the university had at the time. The
only thing that ran 'slow' was the final link line in the build :-)
-Matt
Matthew Dillon
<dillon@backplane.com>
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-arch" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200206280453.g5S4rioB002878>
