reggie@pdn.uucp (George W. Leach) (01/22/88)
[ This looks like real good stuff. Perhaps some of these folks that George ] [ has contacted can be persuaded to contribute a description of their system? ] Here is the summary of the information that was sent in response to my inquiry concerning distributed systems. First, I would like to thank the following individuals who replied: Michel Gien (chorus.fr!mg) Chorus Systemes, Montigny-le-Bretonneux, France Preston Bannister (felix!preston) FileNet Corp., Costa Mesa CA Ken Birman (cornell!ken) Cornell University CS Dept, Ithaca NY (paf@sdcsvax) University of California, San Diego CA Fred Douglis (ginger.berkeley.edu!douglis) University of California, Berkeley CA H. Stephen Au-Yeung (tolerant!stephen) Tolerant Systems, Inc., San Jose CA Dave Wagner (uw-beaver!wagner) University of Washington, Seattle WA Rich A. Hammond (bellcore!hammond) Bell Communications Research, Morristown NJ I would also like to further thank Michel Gien, Ken Birman, Fred Douglis, and H. Stephen Au-Yeung for following up on my further requests for printed matter on their respective systems. Although my original inquiry was directed at acquiring information on Distributed UNIX or UNIX-like systems, I also received at least one reply concerning a network file system. In addition, I received and gathered information on distributed operating systems. Perhaps the wording of my original posting should have been clearer. I was looking for information on distributed operating systems and only meant to use distributed UNIX or UNIX-like as an example! For example, I am also aware of some distributed object-oriented operating systems/environments that have been worked on, eg. Allegro at U. of Stanford or Eden and U. of Washington. ____ ==> [ unzip here ] |____)-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8- Provided below is a summary of various projects or systems that I have compiled. Where the information is know a contact is provided along with an e-mail address is one was available. Project Organization Contact ------------------------------------------------------------------------------- Mach Carnegie-Mellon University Rich F. Rashid DUNIX Bell Communications Research A. Litman DUNE Bellcore/Princeton U. Luis Cova princeton!cova UNIX United U. of Newcastle upon Tyne Brian Randell Newcastle Connection U. of Newcastle upon Tyne Brian Randell Cambridge Ring University of Kent R.P.A. Collinson LOCUS UCLA & Locus Computing Corp. Gerald J. Popek Amoeba Vrije Universiteit Andrew S. Tanenbaum botter!ast V Stanford University D.R. Cheriton Eden U of Wash. (Rice U.) Guy T. Almes rice!almes NEST AT&T BTL, Murray Hill NJ Rakesh Agrawal allegra!rakesh COCANET U of California - Berkeley L.A. Rowe Project Athena MIT/DEC James Gettys jumbo!jg Andrew CMU ITC/IBM James H. Morris NM State Ring New Mexico State University Arthur I. Karshmer MOS Hebrew U of Jerusalem Amnon Barak PULSE University of York G.M. Tomlinson Allegro Stanford University Mark A. Linton lurch.stanford.edu!linton Chorus Chorus Systems Michel Gien chorus!mg FileNet FileNet Corporation Martin McKenery felix!martin ARGUS MIT Barbara Liskov CLOUDS Georgia Tech Richard LeBlanc gatech!rich Sprite U of California - Berkeley Fred Douglis sloth.berkeley.edu!douglis ____ ==> [ unzip here ] |____)-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8- Date: Tue, 29 Dec 87 20:49:07 +0100 From: rutgers!chorus.fr!mg (Michel Gien) Organization: Chorus Systemes, Montigny-le-Bretonneux, France To: moss!codas!pdn!reggie Subject: Re: Distributed UNIX information request Yes, you ought to look into CHORUS....:-) Seriously, Chorus has been a research Project on Distributed Systems at INRIA in France from 1979 to 1986. Going through three iterations, referred to as CHORUS-V0, CHORUS-V1, and CHORUS-V2, all based on a message passing kernel. CHORUS basic concepts for handling distribution are "Ports" and "Messages". System as well as Application services are provided by means of processes (called Actors), communicating by exchanging messages. CHORUS-V2 is UNIX compatible, at the object code level. It supports all standard Unix programs, in a distributed environment. Files, devices, processor resources, etc., distributed over several machines on a LAN, can be accesssed in the same way, and with the same programs, as on a conventional Unix system. CHORUS provide transparent access to "distributed names" that can be associated with any type of service and/or resource, rather than to a specific set of particular remote resources as does NFS for example with files. In december 1986, most of the researchers that had been working on the Chorus research project formed a Company called Chorus syste`mes to further develop, industrialize and market CHORUS on a commercial basis. The current version, being developped by Chorus syste`mes, is called CHORUS-V3. Its main features are: 1. Unix compatibility for user programs is at the EXECUTABLE code level, with other Unix systems running on the same hardware (assuming same compilers and linker), i.e. Unix application executables can be run unchanged on CHORUS-V3. 2. The Unix interface supported includes POSIX, System V (X/OPEN) and 4.3 BSD/SUN extensions (sockets, NFS). 3. Distributed Virtual Memory will be provided with object (e.g. files) mapping into memory over the network. 4. Actors (i.e. processes) can be "multiplexed" into several "activities" (i.e. "threads" ala Mach), sharing the actor's resources. 5. Unix modularity of user programs has been extended to system services: file, process, network management, device drivers, etc. are implemented as "standard" actors, and are not part of the kernel. Therefore, only those system functions actually required need to be installed. They can be complemented or reduced dynamically while the system is running. In particular, drivers can be installed, changed or removed on-line... This provides extremely powerful tools for dynamically optimising distributed system configurations to evolving application requirements. (This is currently supported by CHORUS-V2). 6. The CHORUS kernel itself is implemented as a "multi-threads actor" and "user-level" threads can be dynamically attached to the kernel. 7. User procedures may also be attached dynamically to interrupt vectors to allow efficient handling of interrupts, thus providing tools for real-time applications. 8. Unix compatibility is also provided at drivers level, allowing to port such drivers at very low-cost. 9. Portability of the system has been considerably improved by means of careful study of various hardware architectures, increased modularity of the system components and programming language used for implementation (C++ and C). 10. Quality assurance methods are being used for development of the system software (in C++). Compared to other distributed systems, one could possibly "summarize" (???) CHORUS as a mixture of (best of (!!!)) V-system (message passing kernel), Mach (Virtual Memory and Threads), Amoeba (network addresssing) and Locus (although CHORUS hasn't gone as far yet as regards file replication in particular). And of course Unix compatibility... A first release of CHORUS-V3 is planed to be available for beta-testing 3Q-88, on Bull/SPS-7, SUN-3, and PC/AT-386. I am mailing you a set of papers on CHORUS, most reflecting CHORUS-V2 architecture. I should be able to mail you CHORUS-V3 papers pretty soon. Hope this is usefull information to you. Thanks again for your interest. I am of course very much interested in the other replies you've got from other people working on similar subjects. _ _ _ _ Michel Gien ' ) ) ) / // Chorus systemes / / / o _. /_ _ // 6 avenue Gustave Eiffel / ' (_<_(__/ /_</_</_ F-78180, Montigny-le-Bretonneux (France) Tel: +33 (1) 30 57 00 22 Fax: +33 (1) 30 57 00 66 mg@chorus.fr (Internet) {uunet}!mcvax!inria!chorus!mg (old uucp) ____ ==> [ unzip here ] |____)-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8- Date: Thu, 24 Dec 87 11:55:27 pst From: ihnp4!hplabs!felix!preston (Preston Bannister) To: hplabs!ihnp4!codas!pdn!reggie Subject: Re: Distributed UNIX information request Organization: FileNet Corp., Costa Mesa, CA FileNet uses a version of Unix version 7 with a distributed file system. Hopefully you will get some mail from one of the OS people here. If not then you might try sending mail to edwards or martin on this machine. Login name: edwards In real life: Dave Edwards Project: Operating System Group (OSG) Manager Login name: martin In real life: Martin McKendry The FileNet system is a collection of 680x0 based boxes on an Ethernet all sharing the same filesystem. Our product is a system for storing and using very large numbers of documents. Each box may have a scanner, optical disk jukebox, laser printer, or very high resolution (1700x1200) display attached. The first product shipped over two years ago, so this is not a new development. There are people (Martin in particular) working on improving the performance of the system. -- Preston L. Bannister USENET : ucbvax!trwrb!felix!preston BIX : plb CompuServe : 71350,3505 GEnie : p.bannister ____ ==> [ unzip here ] |____)-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8- Date: Fri, 25 Dec 87 15:17:20 EST From: rutgers!rochester!cornell!ken (Ken Birman) To: moss!codas!pdn!reggie Subject: Re: Distributed UNIX information request Organization: Cornell Univ. CS Dept, Ithaca NY If you include ARGUS as a UNIX-like system, you probably ought to include ISIS (my system), CLOUDS (G. Tech), and perhaps the arizona system, SUGUARO (?sp). Actually, I suspect that neither ARGUS nor V really belong on your list. What are you compiling this information for? Ken Date: Thu, 31 Dec 87 20:26:15 EST From: rutgers!kama.cs.cornell.edu!cornell!ken (Ken Birman) To: moss!codas!pdn!reggie Subject: Re: distributed os. Yes, we do have more written material. As a matter of fact, we plan to release the ISIS system for SUN computers sometime this spring, probably in March. It comes with an extensive manual, and public-domain style support, but might be a good testbed for a project like yours. Performance is remarkably good for something on top of UNIX as long as you use the system for the things that it is good at. I'll have my secretary mail your our current papers, plus a recent one on the RNFS (no current plans to distribute it, I should mention, but this could change). If you are interested in running ISIS on your SUN systems, let me know. Ken Birman ____ ==> [ unzip here ] |____)-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8- Subject: Re: Distributed UNIX information request Date: 26 Dec 87 18:33:20 GMT You might contact Plexus Computers. They developed (and sold) the first comercial Distributed File system in 1982-3 for System-III. This was a rmount model with state information. The designer was Monte Pickard, now at Counterpoint computers. [ This is mostly of historical interest, but I would like to hear more ] [ about it if Monte is somewhere on the net. --DL ] ____ ==> [ unzip here ] |____)-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8- From: Fred Douglis <rutgers!ginger.berkeley.edu!douglis> To: ginger.berkeley.edu!gabato Cc: moss!codas!pdn!reggie, sdcsvax.ucsd.edu!darrell%beowulf Subject: please send Sprite Packet Date: Thu, 24 Dec 87 13:58:19 PST Sender: rutgers!sloth.berkeley.edu!douglis Liza, once you get back, can you please send the packet to: George W. Leach Paradyne Corporation Mail stop LF-207 P.O. Box 2826 Largo, FL 34649-2826 Thanks, - Fred - ____ ==> [ unzip here ] |____)-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8- Date: Thu, 24 Dec 87 08:29:38 pst From: rutgers!ucbvax.berkeley.edu!ames!oliveb!tolerant!stephen (H. Stephen Au-Yeung) To: moss!codas!pdn!reggie Subject: Re: Distributed UNIX information request We, at Tolerant Systems, are developing a fault tolerance distributed UNIX called TX. If you are interested, let me know, I'd see if I can find some printed material to send to you. Stephen Au-Yeung Tolerant Systems, Inc. 81 E. Daggett Drive San Jose, CA. 95134 Date: Sat, 26 Dec 87 22:08:18 pst From: rutgers!pangaea.berkeley.edu!tolerant!stephen (H. Stephen Au-Yeung) To: ames!ames!rutgers.edu!moss!codas!pdn!reggie Subject: TX George, I'll see if I can send you some papers about our OS after this holidays. Yes, TX is designed and implemented by Tolerant Systems and we've sold our source code to several companies all over the world. As I might have told you before, TX is based on 4.2 BSD UNIX and we added the support for on line transaction processing and fault tolerance. To make this possible, some modules of the OS have been rewritten, e.g., the file system. Anyway, wait for more details. Stephen Au-Yeung ____ ==> [ unzip here ] |____)-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8-8- Date: Fri, 25 Dec 87 14:05:44 PST From: usfvax2!gould!uunet!seismo.CSS.GOV!harvard!uw-beaver!uw-june!wagner (Bullwinkle J. Moose) To: gould!novavax!usfvax2!pdn!reggie%seismo.CSS.GOV Subject: Distributed UNIX There is a system called DUNIX that was developed at Bell Communications Research by Ami Litman. I only have a BCR tech report on it, so I don't know if it was ever published anywhere. To my understanding, it's a complete re-write of UNIX, not just a bunch of hacks. The author's address is BCR, 435 South Street, Morristown, New Jersey, 07960. For a reference to a predecessor of DUNIX, see A. Barak and A. Litman, "MOS: A Multicomputer Distributed Operating System", in Sofware Practice and Experience, vol 15(8), pp. 725-737, August, 1985. This work was done at the Hebrew University of Jerusalem. Hope this helps. ___ _ ____ _ ___ / \__/ \__/ \__/ \__/ \ Hey Rocky! | _|@ @ __ | Watch me pull a rabbit \________/ | | \________/ out of my hat! __/ _/ /) (o _/ \____/ Dave Wagner University of Washington Comp Sci Department wagner@cs.washington.edu {ihnp4,harvard,tektronix}!uw-beaver!wagner Date: Thu, 24 Dec 87 11:27:01 est From: rutgers!lafite.bellcore.com!hammond (Rich A. Hammond) To: moss!codas!pdn!reggie Subject: Other Distributed Unix Systems We have 2 distribtued UNIX systems projects here at Bellcore in the applied research area. DUNIX is a distribtued UNIX with an orientation towards availability in the face of processor failures. It is based on large Vaxen (Unibus or BI bus based machines). Dune (internal name, haven't any idea about official name) is oriented towards making use of lots of fast, cheap processors, it uses 68020 based VME/multibus boards. Let me know if this gets through and you want more info. I know the DUNIX stuff is somewhat available, at least to universities. Rich Hammond Bell Communications Research, (201) 829-4383 hammond@bellcore.com