From owner-freebsd-security Sat Feb 28 23:18:52 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id XAA05620 for freebsd-security-outgoing; Sat, 28 Feb 1998 23:18:52 -0800 (PST) (envelope-from owner-freebsd-security@FreeBSD.ORG) Received: from fledge.watson.org (root@FLEDGE.RES.CMU.EDU [128.2.91.116]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id XAA05614 for ; Sat, 28 Feb 1998 23:18:49 -0800 (PST) (envelope-from robert@cyrus.watson.org) Received: from trojanhorse.pr.watson.org (trojanhorse.pr.watson.org [192.0.2.10]) by fledge.watson.org (8.8.8/8.6.10) with SMTP id CAA16116; Sun, 1 Mar 1998 02:18:41 -0500 (EST) Date: Sun, 1 Mar 1998 02:16:09 -0500 (EST) From: Robert Watson X-Sender: robert@trojanhorse.pr.watson.org Reply-To: Robert Watson To: Javier Henderson cc: Christopher J Ceska , freebsd-security@FreeBSD.ORG Subject: Re: Question In-Reply-To: <199803010644.WAA18503@kjsl.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-security@FreeBSD.ORG Precedence: bulk On Sat, 28 Feb 1998, Javier Henderson wrote: > Robert Watson writes: > > > I'm not familiar with the VMS clustering behavior -- what services does it > > provide? > > You can mount disk drives across the clusters. There's a > distributed lock manager, which is what allows the transparent sharing > of resources across the cluster members. What resources? This sounds like it would go beyond just distributed file sharing/management, but given my lack of experience, I hesitate to assume that. When Terry's NFS locking patches are in place, and the client/server side rpc.lockd/statd stuff works, NFS will be able to provide at least some subset of the file-system locking behavior. > Most clusters have a shared authorization file, though you can > have node-specific files as well. Authorization or Authentication (or both? :). I.e., is this more a concept of a distributed master.passwd/group as in NIS or a distributed file system arrangement, or something a little more broad? > You can have multiple hosts and disk drives on the same SCSI > bus, and even if one node crashes, the remaining nodes' access to the > disk drives remains undisturbed in most cases. Any node can also serve > disk drives to the cluster, though of course those drives would go > away if the node crashes. > > That's in a very small nutshell. There's lots more. There has been discussion of this SCSI behavior on FreeBSD-SCSI and elsewhere recently. I believe the conclusion reached was that most easy solutions to the problem are definitely hacks, etc. The shared SCSI bus sounds like an interestin arrangement that would be very nice in a fault-tolerant/highly-available system, but sounds like there would be quite complicated arbitration issues for buses/devices. Are there distributed process facilities to allow farming of processes? A nice LPC/RPC transparency system is always fun. :) I seem to recall a group in Israel was doing some work on something like that on BSD/OS a year or two ago, but I can't remember the specifics. A cool clustering system for a BSD would be very nice to have -- on any of these levels. They tend to involve a lot of expensive research, though, I understand. Robert N Watson Carnegie Mellon University http://www.cmu.edu/ SafePort Network Services http://www.safeport.com/ robert@fledge.watson.org http://www.watson.org/~robert/ To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe security" in the body of the message