Path: utzoo!news-server.csri.toronto.edu!cs.utexas.edu!sdd.hp.com!wuarchive!udel!rochester!cornell!ken
From: ken@gvax.cs.cornell.edu (Ken Birman)
Newsgroups: comp.sys.isis
Subject: Re: ISIS features
Message-ID: <52926@cornell.UUCP>
Date: 8 Mar 91 18:52:53 GMT
References: <1991Mar8.163555.1321@aero.org>
Sender: nobody@cornell.UUCP
Reply-To: ken@gvax.cs.cornell.edu (Ken Birman)
Organization: Cornell Univ. CS Dept, Ithaca NY
Lines: 66

In article <1991Mar8.163555.1321@aero.org> abbott@aero.org (Russell J. Abbott) writes:
>I just read that the Carnot project at MCC plans to offer virtual
>synchrony (plus a whole lot more).  If virtual synchrony becomes one of
>the more-or-less "standard" features that distributed systems begin to
>provide, where will ISIS focus its future efforts?

An answer in 3 parts:

1) I have never heard of the Carnot project and so can't comment on
   this.  Are they implementing the mechanisms themselves?  I would be very
   interesting in seeing the details.  

   There is at least one MCC group that actually uses ISIS in their work.
   Maybe this is that group?

2) We would like our next ISIS system (the one we are designing and building
   now) to be a component of the various distributed systems that are likely
   to be important in the next few years.  To this end:

      - We have been working with the Chorus people on an extension to make
        the Chorus port-group mechanism virtually synchronous.
      - We are working closely with the Mach project on adding
        virtually synchronous port-groups to Mach+Xkernel
      - We are continuing to talk to OSF about our stuff...
      - We are starting to talk to UI about our stuff...
      - We are talking to a number of major hardware vendors about our
        commercial plans.

   So far, this is mostly talk.  However, our new system design and
   implementation is well underway.

   The best outcome from my point of view would be that all these groups
   just buy into ISIS.  I would like to see standards for process group
   mechanisms in UNIX-like systems, and I think that our new ISIS
   architecture could be a very solid basis for such standards.  We have
   much more experience than any other group in this area, and we are making
   the development of the new architecture as "open" a process as possible.

   As an aside, ISIS was not easy to develop, and our publically available
   source is a mixture of old public domain code (pre V1.3) and a lot of
   more recent code subject to a fairly restrictive copyright.  So, any
   group proposing to build a virtually synchronous toolkit has a fair amount      of work cut out for themselves, and they won't be able to use much of
   our recent code (not without our written permission, at any rate).

   However, we built one and are building another, and I am sure that there
   are at least a dozen distributed systems groups world-wide fully capable
   of doing the same.  So, perhaps some real competition will emerge in the
   next few years.  Generally speaking, this would be good...  keeps us
   all honest...  and, my group has probably missed a number of deep
   insights or clever tricks that other groups can contribute, and we all
   benefit when that happens.

3) The focus of our future efforts are:

       - Near term: getting our new lightweight stuff running as solidly and
         as close to the hardware performance limits as possible, and
         re-working the toolkit with a focus on awkward aspects like the
         C++ interface and the way in which the architecture specification
         is presented (i.e., incorporating ideas from projects like ANSA)

       - Longer term: we've been thinking about this.  We'll tell you when we
         get there...  (after all, if we are going to be imitated, we want
         the credit, and you get the credit by getting there 2 years before
         the competition!)

Ken