Encoding Network Addresses to Support Operation over Non-OSI Lower Layers
Extracted elements (20)
A decimal abstract encoding was defined for the DSP rather than a binary ASN.1 structure because there is insufficient space in the Network Address for an ASN.1 encoding to be feasible. ECMA 117 format was also considered but deemed unsuitable.
The new Telex-AFI network address format is explicitly not appropriate for use where CONS or CLNS is used to provide COTS or CLTS; X.121 AFI handles international X.25 cases without PID or CUDF, and the new format covers only the remaining cases.
The Telex AFI was chosen for the IDP of the new network address format because it provides the largest DSP and is less likely than other AFI forms (Local, X.121, DCC, ICD) to be confused with 'real' Network Addresses used for CONS or CLNS.
The two-digit prefix at the start of the DSP allows multiple usages of the same DSP space and permits different encodings within the same IDP, giving flexibility without consuming the entire DSP.
The encoding described in this document covers TCP/IP networks using RFC 1006 and X.25(80) mappings of COTS where X.25 carries only DTE (not Network Address). It is not applicable where CONS or CLNS provides COTS or CLTS.
The International X.25 allocation (Prefix 01 under IDI 007 28722) is only used where a CUDF or PID is needed. In all other international X.25 cases, an X.121 form Network Address with no DSP should be used instead.
All addressing information required for the network (including the SNPA) must be deducible from the Network Address itself, without reliance on a separate directory service at the network layer.
Any End System using the Connection Oriented Network Service with access to the international X.25 service is recommended to use the X.121 form of NSAP address relative to its X.25 access point, to allow routing across worldwide X.25 public data networks.
For the TCP/IP network-specific DSP, there shall be a 20-digit long network-specific part: the IP address occupies the first 12 digits, the optional port number occupies digits 13–17, and the optional transport set occupies digits 18–22.
If an X.121 IDP form Network Address is interpreted, the X.121 address is recommended to provide a route (by defining an SNPA on the international X.25 network), though additional or preferable routes may be determined by private means.
If the DSP is absent in an X.121 AFI network address, the form should be interpreted as implying a mapping of Transport onto X.25(80).
Network Addresses using the new format must be globally unique, as they can be looked up by any party with access to the OSI Directory.
Where an OSI application is run over CLNS on the Internet, the NSAP allocation guidelines of RFC 1237 must be followed.
The DSP top-level structure consists of a 2-digit Prefix followed by a Network Specific part of up to 20 digits (27 for Telex IDI format). The IDP plus the 2-digit prefix together identify a subnetwork in which the remainder of the DSP (Network Specific Part) is interpreted.
The Transport Set field in the TCP/IP DSP format is a 16-bit flag word encoded as a 5-digit decimal number. Value 1 means TCP, value 2 means UDP; further values are assigned by IANA. Absence or all-zero bits means 'default' (TCP).
The X.25(80) DSP Type field governs the presence of optional PID/CUDF: Type 0 encodes DTE only; Type 1 encodes DTE plus PID; Type 2 encodes DTE plus CUDF; Types 3–9 are reserved.
Appendix A allocates well-known Telex-AFI network address prefixes under IDI 007 28722: Prefix 01 for International X.25, Prefix 02 for Janet, Prefix 03 for DARPA/NSF Internet (delegated to IANA), and Prefix 06 for IXI. These allocations are committed as stable by UCL.
Security considerations are explicitly not discussed in this memo.
The TCP/IP (RFC 1006) network-specific DSP format is exactly 20 digits: digits 1–12 encode the IP address (4 octets, each as 3 decimal digits), digits 13–17 encode the optional port number (5 digits, defaulting to the RFC 1006 port), and digits 18–22 encode the optional transport set (5 digits; 1=TCP, 2=UDP; absent or zero means default TCP).
The X.25(80) network-specific DSP format encodes a 1-digit Type field (0=DTE only, 1=DTE+PID, 2=DTE+CUDF, 3-9 reserved), followed when Type>0 by a 1-digit PID/CUDF Length (n octets), then 3n decimal digits for PID/CUDF (each octet encoded as three decimal digits), and finally the DTE number occupying the remainder of the address.