Browse Prior Art Database

Proposed User-User Protocol (RFC0091) Disclosure Number: IPCOM000003960D
Original Publication Date: 1970-Dec-27
Included in the Prior Art Database: 2001-Oct-17
Document File: 13 page(s) / 27K

Publishing Venue

Internet Society Requests For Comment (RFCs)

Related People

G.H. Mealy: AUTHOR



This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 13% of the total text.

Network Working Group                                    George H. Mealy

Request for Comments: 91                              Harvard University

                                                       December 27, 1970

                     A PROPOSED USER-USER PROTOCOL


   There are many good reasons, and maybe one or two bad ones, for

   making it appear that communication over the Network is only a

   special case of input/output -- at least as far as user programming

   is concerned.  Thus, for instance, the Harvard approach toward

   implementing the HOST-HOST protocol and Network Control Program

   treats each link as a "logical device" in PDP-10 terminology.

   Setting up a connection is similar to local device assignment, and

   communication over a link will make use of the standard system

   input/output UUO's.  This makes it possible to use existing programs

   in conjunction with the Network without modification -- at least if

   other PDP-10's are being dealt with.

   This takes us only so far, however.  The notion of a "logical device"

   does not exist on the PDP-10; it does on the IBM 360 (I am speaking

   here at the level of the operating system -- user program interface).

   Furthermore, in the absence of a Network standard requiring fixed

   representations for integers, reals, etc. (which I would oppose), any

   pair of user processes must arrive at a local agreement, and one or

   both must assume the burden of data conversion where necessary.  Any

   standard protocol should allow such agreements to be given expression

   and should accommodate at least the minimum of control information

   that will allow such agreements to function in practice.  Finally, we

   must note that the IMP-IMP and HOST-HOST protocols do not provide for

   a check that an action requested by a user process is actually

   accomplished by the other processes; this type of issue has always

   been regarded as subject to treatment at the USER-USER protocol


   This proposal is intended to face the above three types of issue only

   to a certain extent.  I can best explain that extent by stating the

   criteria I would use to judge any USER-USER protocol proposal:

Mealy                                                           [Page 1]

RFC 91               A Proposed User-User Protocol         December 1970

   1.   The notion of a (logical) _record_ should be present, and the

        notion of a _message_ should be suppressed. (To a FORTRAN pro-

        grammer, that which is written using one WRITE statement with no

        accompanying FORMAT is a record; to an OS/360 machine language

        programmer, PUT writes a record).

   2.   It should be possible to so implement the protocol in HOST sys-

        tems and/or library routines that now existing user programs can

        access files anywhere in the Network without program modifica-

        tion. (Initially, at least, this ability must be restricted to

        HOST systems of the same type).

   3.   The protocol should be implementable (not necessarily imple-

        mented) in any HOST system at the SVC or UUO level.  Specific

        knowledge of the characteristics of the other HOST involved

        should be unnecessary.

   It should be noted that the above imply that some user programs must

   be aware of the nature of the other HOST -- at least in each case

   where the second crit...