rfc1007.txt

著名的RFC文档,其中有一些文档是已经翻译成中文的的.

     a.   TSAP identifier(s): the parameters values shall follow the
          conventions given in Paragraphs 5.1.1 and 5.1.2.

     b.   TPDU size: (The values for this parameter are under
          consideration.)

     c.   Version number: as in IS 8073.

     d.   Protection parameters: should not be used.

     e.   Checksum: if Class 4 has been selected, this parameter may be
          used. If Class 2 (or Class 0) has been selected, this
          parameter shall not be used.

     f.   Additional options: if Class 4 or Class 2 has been selected,
          this field shall take binary value 0001 if the responding user
          has proposed use of expedited data and shall take binary value
          0000 otherwise.

     g.   Alternate protocol classes: shall not be used.

     h.   Acknowledgement time: should not be used.

     i.   Throughput: should not be used.

     j.   Residual error rate: (The use of this parameter is under
          consideration.)

     k.   Priority: (The use of this parameter is under consideration.)

     l.   Transit delay: should not be used.

4.4.2.2.1.3   User data.

   The CC TPDU shall not carry user data which has any requirement
   whatever for security protection.




McCoy                                                          [Page 18]

RFC 1007                                                       June 1987


4.5   Use of local options.

   The paragraphs that follow give policy and guidance in the election
   of local options.

4.5.1   Local options, Class 2.

4.5.1.1   Multiplexing.

   Any Class 2 connections may be multiplexed on the same network
   connection to the limits provided by the network service.
   Multiplexing Class 2 and Class 4 connections together on the same
   network connection is not recommended.

4.5.1.2   Acknowledgement strategy.

   (The provisions of this paragraph are under consideration.)

4.5.1.3   Concatenation.

   This permits placing certain TPDUs into a single network service
   data unit with a data-bearing TPDU.  It is useful for reducing
   the overhead of separate transmission of the individual TPDUs.

4.5.1.4   Lockup prevention timer.

   It is strongly recommended that this timer be used for all Class 2
   connections.  A description of the timer has been included in the
   transport formal description.  (This timer corresponds to the
   optional TS1 timer that IS 8073 recommends.)

4.5.1.5   Treatment of protocol errors.

   Protocol errors detected by a Class 2 transport connection shall
   result in that connection being terminated, without sending an ER
   TPDU.

4.5.1.6   Action on receipt of Error transport protocol data unit.

   The receipt of an ER TPDU for a Class 2 transport connection shall
   cause immediate termination of that transport connection.

4.5.2   Local options, Class 4.

4.5.2.1   Withdrawal of flow control credit.

   Because of the need to serve transport connections of various
   levels of operating priority, an implementation shall support
   the withdrawal of flow control credit from any Class 4 transport
   connection as a means of managing resource allocation among
   Class 4 connections.



McCoy                                                          [Page 19]

RFC 1007                                                       June 1987


4.5.2.2   Flow control confirmation.

   The requirement to support withdrawal of flow control credit
   strongly indicates the need to use flow control confirmation.
   An implementation should support and use the flow control
   confirmation procedures of IS 8073, consistent with quality of
   service and other requirements.

4.5.2.3   Subsequenced acknowledgements.

   The possibility of credit withdrawal strongly indicates the
   requirement for subsequence numbers on acknowledgements.  An
   implementation shall support and use subsequence numbers as
   defined in IS 8073.

4.5.2.4   Splitting over network connection.

   Implementations may use splitting as necessary or useful in the
   operating environment. (Splitting is defined only for operation
   over a CONS.

4.5.2.5   Acknowledgement strategy.

   (The provisions of this paragraph are under consideration.)

4.5.2.6   Wait-before-closing state.

   It is recommended that this state be used.  A lockup prevention
   timer, such as used in Class 2, is not necessary, since the CR
   TPDU retransmission timer serves this purpose.

4.5.2.7   Multiplexing on network connection.

   Multiplexing of Class 4 connections on a single network
   connection may be used as necessary or useful, within the limits
   permitted by the network service.  Class 4 connections should not
   be multiplexed onto network connections serving Class 2 transport
   connections.

4.5.2.8   Concatenation of protocol data units.

   Concatenation may be useful when operating over a CLNS that has
   large capacity service data units.  Concatenation on networks
   that areconnection-oriented may be useful if transport
   connections are being multiplexed.  A careful analysis of the
   treatment of the network service data unit in internetwork
   environments should be done to determine whether concatenation
   of TPDUs provides sufficient benefit to justify its usage in
   those circumstances.





McCoy                                                          [Page 20]

RFC 1007                                                       June 1987


4.5.2.9   Checksum algorithm.

   It is strongly recommended that the algorithm described in the
   Implementors Guide Part 7, be used rather than the algorithm
   given in the Annex to IS 8073.  The algorithm in  Part 7
   computes the same checksum as the one in IS 8073 but  has been
   optimized.  Guidance on the use and non-use of  checksum is
   given in the Implementors Guide, Part 7.

4.5.2.10   Send network RESET when bad checksum received.


   It is recommended that only an N-RESET be sent when encountering
   a TPDU with a bad checksum on a CONS.  An implementation shall
   not send an N-DISCONNECT-request in such situations, since the
   TPDU with the bad checksum may have come from some entity
   intending to interfere with communications.  When  operating
   Class 4 over a CLNS, no action shall be taken on the receipt of
   a TPDU with a bad checksum,  i.e., the TPDU shall be discarded.

4.5.2.11   Protocol data unit retransmission policy.

   (The provisions of this paragraph are under consideration.)

4.5.2.12   Treatment of protocol errors.

   In Class 4, a protocol error arising from a TPDU containing
   unrecognized parameters shall cause a DR TPDU to be sent to the
   sender, if the TPDU is otherwise valid.  All other erroneous TPDUs
   shall be discarded.

4.5.2.13   Action on receipt of Error transport protocol data unit.

   If an ER TPDU is received from a remote transport entity, an
   implementation to which this supplement applies shall release the
   transport connection with which the ER TPDU is associated, if such
   association can be made.  When association cannot be made, the ER
   TPDU shall be discarded.

5. SPECIAL REQUIREMENTS

5.1   Addressing conventions.

   (The provisions of Paragraph 5.1 and its subparagraphs are under
   consideration.)

5.1.1   Transport Service Access Point.

5.1.2   Connect-request/confirm protocol data units.

5.1.3   Network Service Access Point.



McCoy                                                          [Page 21]

RFC 1007                                                       June 1987


5.2   Convention for use of transport connection reference numbers.

   The ISO Transport Protocol provides for freezing reference numbers
   by means of a timer, so that re-use of a reference number does not
   cause ambiguity in communications. However, certain requirements
   are imposed on DOD implementations, so that this means of reference
   number control is inadequate alone. The ISO standard defines only
   those actions to be followed if a timer is used.  Other means of
   reference number control are not prohibited, providing that the
   minimum freeze time, as defined in IS 8073, is exceeded for each
   reference number used.

5.2.1   Specification of convention.

   An implementation adhering to the applications definitions in
   this supplement, Paragraph 1.3, shall not re-use a transport
   connection reference number until the set of available reference
   numbers has recycled to that point.  Expressed more formally,
   if all reference numbers are defined to be within the interval
   [1,N] and a reference number R in this interval is used, then
   R shall be prohibited from being selected again until all the
   numbers R+1,...,N,1,2,...,R-1 shall have been used.  The choice
   of N should be sufficiently large that the expected recycle period
   exceeds the minimum freeze time as specified in IS 8073.  This
   requirement is in addition to and does not supersede the freeze
   requirement of IS 8073.  A simple means of implementing this
   convention is given in Part 9.3 of the Implementors Guide.

5.3  Operation over connectionless network service.

   Implementations to which this supplement applies are required to
   operate over connectionless network services in addition to being
   able to operate over connection-oriented network services. The ISO
   standard specifies transport only for operation over a
   connection-oriented network.  However, the specification for Class
   4 has been written in such a way that use with connectionless
   network service is not precluded.  The formal description offers
   even more flexibility in this regard.  Consequently, operation over
   connectionless network services, whether a LAN or IP, is primarily
   an implementation issue for Class 4.  Operation of Class 2
   transportover a connectionless network service is not considered
   to be a reasonable  option because of the lack of sufficent error
   recovery in Class 2.   For the purposes of this supplement,
   operation of Class 2 on a  connectionless network service is
   not recommended. Operation of Class  4  over a connectionless
   network service is discussed further in parts 1.2.2.2,  3.4,
   and 6 of the accompanying Implementors Guide.







McCoy                                                          [Page 22]

RFC 1007                                                       June 1987


5.4   Recovery from peer deactivation.

   The ISO Standard does not provide for re-establishment of the
   transport connection  when one of the communicating peers is
   deactivated ("crashes").  However, the state tables for Class
   4 transport in Annex A to IS 8073  are flexible enough that
   simple adaptations in an implementation can yield some degree
   of crash recovery without change to the protocol.  These
   adaptations are discussed in Part 9.2 of the Implementors Guide.













































McCoy                                                          [Page 23]