Network Working Group                                      C. Partridge
Request for Comments: 1151                 BBN Systems and Technologies
Updates: RFC 908                                              R. Hinden
                                              BBN Communications Corp.
                                                            April 1990


            Version 2 of the Reliable Data Protocol (RDP)

Status of this Memo

  This RFC suggests several updates to the specification of the
  Reliable Data Protocol (RDP) in RFC-908 based on experience with the
  protocol.  This revised version of the protocol is experimental.

  Distribution of this memo is unlimited.

Introduction

  Experiments in 1986 and 1987 turned up some ambiguities and problems
  with the RDP specification.  At the time, it was hoped that the
  authors might find the time to revise the entire RDP specification to
  fix these problems, however given the limited demand for RDP
  implementations, the authors were never able to justify the time
  involved in revising the spec.  This document lists the changes that
  we believe are appropriate to make to RDP version 1.

  Readers are expected to be familiar with RFC-908.

Changes To The Protocol Header

  There are three changes to the protocol header: the checksum
  algorithm has been changed, the port size increased, and the version
  number incremented.  The new header format is shown in Figure 1.

  The major discovery during the testing of the protocol is that cost
  of computing the the RDP checksum proved surprisingly variable; its
  performance was more heavily affected by the host's data
  representation than anticipated.  Optimized checksum implementations
  on two comparable hardware bases gave performance that differed by a
  factor of five.  Since the speed of the checksum is a key factor in
  the performance of the protocol itself, this variation caused a
  noticeable difference in throughput.

  The wide variation in performance on comparable machines was felt to
  be undesirable, so the checksum has been changed.  RDP now uses the
  16-bit TCP checksum, which is specified on page 16 of RFC-793.




Partridge & Hinden                                              [Page 1]

RFC 1151                    RDP - Version 2                   April 1990


  The 8-bit port size is probably too small to support a large range of
  applications.  Accordingly, the port size is now 16-bits.  Port
  numbers less than 1024 are reserved for well-defined applications.
  Allocable ports begin at port number 1024.

  Finally, because the checksum and port size have been changed, the
  version number has been increased to 2.


                  0             0 0   1         1
                  0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5
                 +-+-+-+-+-+-+---+---------------+
                 |S|A|E|R|N| |Ver|    Header     |
               0 |Y|C|A|S|U|0|No.|    Length     |
                 |N|K|K|T|L| |   |               |
                 +-+-+-+-+-+-+---+---------------+
               1 |         Source Port           |
                 +---------------+---------------+
               2 |       Destination Port        |
                 +---------------+---------------+
               3 |          Data  Length         |
                 +---------------+---------------+
               4 |                               |
                 +---    Sequence Number      ---+
               5 |                               |
                 +---------------+---------------+
               6 |                               |
                 +--- Acknowledgement Number  ---+
               7 |                               |
                 +---------------+---------------+
               8 |           Checksum            |
                 +---------------+---------------+
               9 |     Variable Header Area      |
                 .                               .
                 .                               .
                 |                               |
                 +---------------+---------------+

                        RDP Header Format
                            Figure 1

Minor Errors and Ambiguities

  Some ambiguities and minor errors have been found in RFC-908.  They
  are corrected in this section.

  The value of the state variable, SND.UNA is treated inconsistently in
  the pseudo-code on pages 21-29.  On page 12, SND.UNA is defined as



Partridge & Hinden                                              [Page 2]

RFC 1151                    RDP - Version 2                   April 1990


  "the sequence number of the oldest unacknowledged segment", and on
  page 21 it is appropriately set to the initial sequence number when
  the connection is opened.  But on page 28, when an acknowledgement is
  received, SND.UNA is set to SEG.ACK, the sequence number being
  acknowledged, instead of SEG.ACK+1.  A similar inconsistency occurs
  on page 26.  The proper fix is to always set SND.UNA to SEG.ACK+1.

  The pseudo-code on page 25 for the SYN-SENT state is incorrect.  The
  first few lines cause all packets with the ACK bit set to be
  discarded, but later lines test the ACK bit.  The test for the ACK
  bit should be placed after all the other tests.  Also note that if
  the ACK bit is set, a RST segment is sent to reset the remote peer,
  but the local peer is left half-open.  There is a similar problem in
  the SYN-RCVD state.  The local peer should deallocate the connection
  record and close.

  On page 24, the pseudo-code indicates that if non-data packets are
  received in the CLOSED state, a RST segment with SEG.ACK set to
  RCV.CUR should be sent.  RCV.CUR is not defined in the CLOSED state.
  SEG.ACK should be set to SEG.SEG.

  There is some inconsistency about how to handle a RST packet in the
  CLOSE-WAIT state.  On page 24, the pseudo-code shows that a RST
  should cause the connection state to become CLOSED.  Text on page 13
  and the state diagram on page 10 suggest the connection state should
  stay in CLOSE-WAIT.  The implementation should stay in CLOSE-WAIT.

  On page 29, the pseudo-code for the OPEN state suggests that if a
  data packet is received in sequence, the acknowledgement packet
  should not contain EACKs.  This is misleading.  Implementations may
  include EACKs in the acknowledgement.

  On page 18, it is possible to interpret the right edge as being
  either inside or outside the window.  This results in a one segment
  difference in the window size.  The proper interpretation is that the
  right edge is outside the window.  In other words, the right edge is
  the first sequence number that cannot be sent or received and the
  total window size is 2*X, where X is the maximum number of
  outstanding segments.

  One final problem is that RDP's flow control scheme does not allow
  the receiver to close the sender's window.  As a result, if the
  receiver acknowledges segments when they are received the sender can
  easily send more data than the receiver is prepared to buffer.  A
  solution to this problem (suggested by members of the End-2-End
  Research Group) is to only acknowledge a segment after it has been
  delivered to the application.  This scheme, however, has not be
  tested.



Partridge & Hinden                                              [Page 3]

RFC 1151                    RDP - Version 2                   April 1990


Security Considerations

  Security issues are not addressed in this memo.

Authors' Addresses

  Craig Partridge
  Bolt Beranek and Newman Inc.
  50 Moulton Street
  Cambridge, MA 02138

  Phone: (617) 873-2459

  EMail: [email protected]


  Robert Hinden
  Bolt Beranek and Newman Inc.
  50 Moulton Street
  Cambridge, MA 02238

  Phone: (617) 873-3757

  Email: [email protected]



























Partridge & Hinden                                              [Page 4]