Message-Id: <199507270530.WAA03778@faerie.CS.Berkeley.EDU>
From: aoki@CS.Berkeley.EDU (Paul M. Aoki)
To: zzhang@csrd.uiuc.edu (Z. Zhang (BambooMan))
cc: postgres-arch@postgres.Berkeley.EDU
Subject: Re: help on XPRS 
Reply-To: aoki@CS.Berkeley.EDU (Paul M. Aoki)
In-reply-to: Your message of Wed, 26 Jul 1995 17:24:01 -0500 (CDT) 
	     <9507262224.AA23230@sp91.csrd.uiuc.edu> 
Date: Wed, 26 Jul 95 22:30:35 -0700
Sender: aoki@postgres.Berkeley.EDU

zzhang@csrd.uiuc.edu (Z. Zhang (BambooMan)) writes:
> Some other fellows pointed to me that XPRS may probably be the only
> parallel database out there that assumes a shared-memory architecture,
> is that true? If so could you point me to (or better yet, ftp site) archieves
> of any tech reports/postscript papers on that project?

commercial:
i believe most systems support global buffer pools for multiple
server processes on shared memory machines

research:
lots of shared memory systems
actual volcano implementations were all shared-memory
laurel at waterloo was shared-memory
i think prisma was too
kardamom somewhere in germany

xprs: 
see http://s2k-ftp.cs.berkeley.edu:8000/postgres/papers
(server machine is down right now, try in a couple of days)

xprs was implemented using postgres on a sequent
most xprs code is still in postgres 4.2 release
not *too* hard to fix up and port to a new platform (took me
a week or so to run on our alpha axp alphaserver, mostly
issues with getting mutual exclusion right on a weak consistency
memory model arch)
--
  Paul M. Aoki          |  University of California at Berkeley
  aoki@CS.Berkeley.EDU  |  Dept. of EECS, Computer Science Division (#1776) 
                        |  Berkeley, CA 94720-1776