CLNP for TUBA
Dave Piscitello <dave@sabre.bellcore.com> Mon, 10 August 1992 21:37 UTC
Received: from ietf.nri.reston.va.us by IETF.NRI.Reston.VA.US id aa20047; 10 Aug 92 17:37 EDT
Received: from NRI.NRI.Reston.Va.US by IETF.NRI.Reston.VA.US id aa20041; 10 Aug 92 17:37 EDT
Received: from p.lanl.gov by NRI.Reston.VA.US id aa12559; 10 Aug 92 17:38 EDT
Received: from noc-gw.lanl.gov by p.lanl.gov (5.65/1.14) id AA08127; Mon, 10 Aug 92 15:30:20 -0600
Received: by noc-gw.lanl.gov (4.1/SMI-4.1) id AA06215; Mon, 10 Aug 92 15:25:48 MDT
Received: from p.lanl.gov by noc-gw.lanl.gov (4.1/SMI-4.1) id AA06211; Mon, 10 Aug 92 15:25:45 MDT
Received: from sabre.bellcore.com by p.lanl.gov (5.65/1.14) id AA07951; Mon, 10 Aug 92 15:25:43 -0600
Received: by sabre.bellcore.com (5.57/Ultrix2.4-C) id AA00193; Mon, 10 Aug 92 16:21:33 EDT
Return-Path: <dave@sabre.bellcore.com>
Received: by japan (4.1/4.7) id AA03310; Mon, 10 Aug 92 16:03:19 EDT
Date: Mon, 10 Aug 1992 16:03:19 -0400
From: Dave Piscitello <dave@sabre.bellcore.com>
X-Station-Sent-From: japan.bellcore.com
Message-Id: <9208102003.AA03310@japan>
To: tuba@lanl.gov
Subject: CLNP for TUBA
Hi, I've taken a cut at "profiling" ISO 8473 as it might be used in TUBA environments. I've tried to respond to many of the criticisms of CLNP posted on big-internets (there are so many I am sure I missed at least one:-) I specifically did not address the issue of the checksum; as Ross has indicated in earlier postings, the issue of whether the checksum is too slow or more powerful is one I don't have sufficient "field experience" information to evaluate; opinions as to how this should be handled are solicited. I don't think I've done anything to CLNP that would make it incompatible with existing implementations; let me know if I have:-) --------- IETF Page 1 August 10, 1992 CLNP for TUBA Internet Draft Use of ISO CLNP in TUBA Environments Dave Piscitello Bellcore dave@sabre.bellcore.com Status of this Memo This document specifies a profile of the ISO 8473 Connectionless-mode Network Layer Protocol data units (CLNP, [1]) for use in conjunction with RFC 1347, TCP/UDP over Bigger Addresses (TUBA, [2]). This draft document will be submitted to the RFC editor as a protocol specification. Distribution of this memo is unlimited. Please send comments to dave@sabre.bellcore.com. Abstract This document describes the use of CLNP to provide the lower- level service expected by Transmission Control Protocol (TCP, [3]) and User Datagram Protocol (UDP, [4]). CLNP provides essentially the same datagram service as Internet Protocol (IP, [5]), but offers a means of conveying bigger network addresses (with additional structure, in order to aid routing). While the protocols offer nearly the same services, IP and CLNP are not identical. This document describes a means of preserving the semantics of IP information that is absent from CLNP while preserving consistency between the use of CLNP in Internet and OSI environments. This maximizes the use of already-deployed CLNP implementations. Acknowledgments Many thanks to Ross Callon of Digital Equipment and Dave Katz of Cisco Systems for their lightning-quick replies to my frequent requests for sanity-checks. Conventions The following language conventions are used in the items of specification in this document: o+ Must, Shall, or Mandatory -- the item is an absolute requirement of the specification. o+ Should or Recommended -- the item should generally be followed for all but exceptional circumstances. IETF Page 2 Internet Draft CLNP for TUBA August 10, 1992 o+ May or Optional -- the item is truly optional and may be followed or ignored according to the needs of the implementor. 1. Terminology To the extent possible, this document is written in the language of the Internet. For example, packet is used rather than "protocol data unit"; "fragment" is used rather than "segment" and byte is used rather than "octet". There are some terms that carry over from OSI; these are, for the most part, used so that cross-reference between this document and RFC994 or ISO 8473 is not entirely painful. OSI acronyms are for the most part avoided. 2. Introduction The goal of this specification is to allow compatible and interoperable implementations to encapsulate TCP and UDP packets in CLNP data units. It is assumed that readers are familiar with RFC 791 and ISO 8473. This document is compatible with (although more restrictive than) ISO 8473. However, it is intended that this document be able to stand on its own without reference to ISO 8473. [Editor's Note: RFC 994 contains the Draft International Standard version of ISO CLNP [6]; while this is not the final version of the ISO protocol specification, it should provide sufficient background for the purpose of understanding the relationship of CLNP to IP, and the means whereby IP information is to be encoded in CLNP header fields.] 3. Overview of CLNP ISO CLNP is a datagram network protocol. It provides fundamentally the same underlying service to a transport layer as IP. Table 1 provides a high-level comparison of CLNP to IP: IETF Page 3 August 10, 1992 CLNP for TUBA Internet Draft Function | ISO CLNP | DOD IP ------------------------|-----------------------|----------------- Version Identifier | 1 byte | 4 bits Header Length | indicated in bytes | in 32-bit words Total Length | 16 bits, in bytes | 16 bits, in bytes Data Unit Identifier | 16 bits | 16 bits Flags | Fragmentation allowed,| Don't Fragment, | More Fragments | More Fragments, | Suppress Error Reports| <not defined> Fragment offset | 16 bits, in bytes | 13 bits, 8-byte units Lifetime (Time to live) | 500 msec units | 1 sec units Higher Layer Protocol | Selector in address | PROTOcol (assigned #) Header Checksum | 16-bit (Fletcher) | 16-bit Addressing | Variable length | 32-bit fixed Options | Security | Security | Priority | <not defined> | Complete Source Route | Strict Source Route | Quality of Service | Type of Service | Partial Source Route | Loose Source Route | Record Route | Record Route | Padding | Padding | <not defined> | Timestamp Table 1. Comparison of IP to CLNP Note that the encoding of options differs between the two protocols, as do the means of higher level protocol identification. Note also that CLNP and IP differ in the way header and fragment lengths are represented, and that the granularity of lifetime control (time-to-live) is finer in CLNP. Some of these differences are non-issues, as CLNP provides flexibility in the way that certain options may be specified and encoded (this will facilitate the use and encoding of certain IP options without change in syntax); others, e.g., higher level protocol identification, must be accommodated in a transparent manner in this profile for correct operation of TCP and UDP, and continued interoperability with OSI implementations. Section 4 describes how header fields of CLNP must be populated to satisfy the needs of TCP and UDP. CLNP and IP also differ in the way in which errors are reported to hosts. In IP environments, the Internet Control Message Protocol (ICMP, [7]) is used to return (error) messages to hosts that originate packets that cannot be processed. An unique message type, the Error Report, is used in CLNP environments. Table 2 provides a loose comparison of ICMP messages to CLNP Error Reports. IETF Page 4 Internet Draft CLNP for TUBA August 10, 1992 CLNP Error Report | ICMP Message --------------------------------|--------------------------------- Reason not specified | <> Protocol Procedure Error | <> Incorrect Checksum | <> PDU Discarded--Congestion | Source Quench (Note 1) Header Syntax Error | Parameter problem Needed to Fragment, could not | Needed to Fragment, Don't Fragment set Incomplete PDU received | <> Duplicate Option | <> Destination Unreachable | Destination Unreachable Destination Unknown | Destination Unreachable Source Routing Error | Source Route failed Unknown Address in Source Route | Source Route failed(?) Path not acceptable | <> Lifetime expired in transit | Time to live exceeded Reassembly Lifetime Expired | Fragment reassembly time exceeded Unsupported Option | <> Unsupported Protocol Version | Parameter problem Unsupported Security Option | Parameter problem Unsupported Source Route Option | Parameter problem Unsupported Record Route option | Parameter problem Reassembly interference | <> [Editor's Note: <> means I could find not comparable value; it is an open issue whether these error messages may be used in TUBA environments. Some of these error messages may correspond to ICMP messages described in RFC1122, Host Requirements. A complete analysis of the error reporting of CLNP and ICMP is required.] Note 1: The current use of the source quench is only when packets are discarded, and thus the current use meaning is the same; if a future RFC describes a more robust treatment of the source quench, the applicability of this CLNP Error Report Type should be reconsidered. Table 2. Comparison of CLNP Error Reports to ICMP Messages 4. CLNP Header Format--Internet Style A summary of the contents of the CLNP header, as it is proposed for use in TUBA environments follows: IETF Page 5 August 10, 1992 CLNP for TUBA Internet Draft 0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | ........Data Link Header........ | NLP ID | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |Header Length | Version | Lifetime (TTL)|Flags| Type | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Fragment Length | Checksum | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Dest Addr Len | Destination Address... | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | ... Destination Address... | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | ... Destination Address... | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | ... Destination Address... | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | ... Destination Address... | PROTO field | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Src Addr Len | Source Address... | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | ... Source Address... | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | ... Source Address... | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | ... Source Address... | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | ... Source Address... | Reserved | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Data Unit Identifier | Fragment Offset | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Total Length of Packet | Options... | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | | : : | Options (see Table 1) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Note that each tick mark represents one bit position. Figure 1. CLNP for TUBA Due to differences in link layer protocols, it is not possible to ensure that the packet starts on an even alignment. Note, however, that many links happen to use a odd length link (eg, 802.2). Thus, the Network Layer Protocol Identifier, although nominally the first byte of the CLNP header, may be considered as part of the link layer in order to re-obtain even alignment. As profiled, the rest of the CLNP packet is even aligned. IETF Page 6 Internet Draft CLNP for TUBA August 10, 1992 The encoding of CLNP fields for use in TUBA environments is as follows. 4.1 Network Layer Protocol Identification (NLP ID) This one-byte field identifies this as the ISO 8473 protocol; it must set to binary 1000 0001. 4.2 Header Length Indication (Header Length) Header Length is the length of the CLNP header in bytes, and thus points to the beginning of the data. Note that the minimum value for a correct header is 55 (assumes source and destination addresses are always precisely 19 bytes long, including the "protocol" and "protocol/reserved" fields, respectively. Also assumes no options are present). The value 255 is reserved. The header length is the same for all fragments of an originated CLNP packet. 4.3 Version This one-byte field identifies the version of the protocol; it is set to a binary value 0000 0001. 4.4 Lifetime (TTL) Like the TTL field of IP, this field indicates the maximum time the datagram is allowed to remain in the internet system. If this field contains the value zero, then the datagram must be destroyed. This field is modified in internet header processing. The time is measured in units of 500 milliseconds, but since every module that processes a datagram must decrease the TTL by at least one even if it process the datagram in less than a second, the TTL must be thought of only as an upper bound on the time a datagram may exist. The intention is to cause undeliverable datagrams to be discarded, and to bound the maximum CLNP datagram lifetime. 4.5 Flags Three flags are defined. These occupy bits 0, 1, and 2 of the Flags/Type byte: 0 1 2 +---+---+---+ | F | M | E | | P | F | R | +---+---+---+ The Fragmentation Permitted (FP) flag, when set to a value of one (1) is semantically equivalent to the "may fragment" value of the Don't Fragment field of IP; similarly, when set to zero (0), the Fragmentation Permitted flag is semantically equivalent to the IETF Page 7 August 10, 1992 CLNP for TUBA Internet Draft "Don't Fragment" value of the Don't Fragment Flag of IP. If the Fragmentation Permitted field has the value O, then the Data Unit Identifier, Fragment Offset, and Total Length fields are not present. The More Fragments flag of CLNP is semantically and syntactically the same as the More Fragments flag of IP; a value of one (1) indicates that more segments/fragments are forthcoming; a value of zero (0) indicates that the last byte of the original packet is present in this segment. The Error Report (ER) flag is used to suppress the generation of an error message by a host/router that detects an error during the processing of a CLNP datagram; a value of one (1) indicates that the host that originated this datagram thinks error reports are useful, and would dearly love to receive one if a host/router finds it necessary to discard its datagram(s). 4.6 Type field The type field distinguishes data CLNP packets from Error Reports; a value of binary 11100 in bits 3-7 of the Flags/Type byte indicates that this is a data packet; a value of binary 00001 (go figure...) indicates that this is an Error Report. 0 1 2 3 4 5 6 7 +---+---+---+---+---+---+---+---+ | flags | 1 | 1 | 1 | 0 | 0 | => Encoding of Type = data packet +---+---+---+---+---+---+---+---+ | flags | 0 | 0 | 0 | 0 | 0 | => Encoding of Type = error report +---+---+---+---+---+---+---+---+ 4.7 Fragment Length Like the Total Length of the IP header, the Fragment length field contains the length in bytes of the fragment (i.e., this datagram) including both header and data. [Note: CLNP also has a Total Length field, that contains the length of the original data submitted by the higher level protocol, e.g., TCP or UDP). 4.8 Checksum A checksum on the header only, verified at each host/router that processes the packet; if header fields are changed during processing (e.g., the Lifetime), the checksum is modified. If the checksum is not used, this field must be coded with a value of zero (0). See Annex A. IETF Page 8 Internet Draft CLNP for TUBA August 10, 1992 4.9 Destination Address Length Indicator () This field indicates the length, in bytes, of the Destination Address. It must be set to the value 19. 4.10 Destination Address The format of the address encoded in this field is described in a companion addressing document, see [8]. For compatibility and interoperability with OSI use of CLNP, the initial byte of the Destination Address is assumed to be an Authority and Format Indicator, as defined in ISO 8348 [7]. A destination address must be 19 bytes long; the 19th byte must always contain the value of the PROTO field, as defined in IP. The 8-bit PROTO field indicates the next level protocol used in the data portion of the CLNP datagram. The values for various protocols are specified in "Assigned Numbers" [9]. 4.11 Source Address Length Indicator () This field indicates the length, in bytes, of the Source Address. It must be set to the value 19. 4.12 Source Address The format of the address encoded in this field is described in a companion addressing document, see [8]. For compatibility and interoperability with OSI use of CLNP, the initial byte of the Destination Address is assumed to be an Authority and Format Indicator, as defined in ISO 8348 [7]. A destination address must be 19 bytes long; the 19th byte must always be reserved. It may be set to the protocol field value on transmission, and shall be ignored on reception. 4.13 Data Unit Identifier Like the Identification field of IP, this 16-bit field is used to distinguish segments of the same (original) packet for the purposes of reassembly. 4.14 Fragment Offset Like the Fragment Offset of IP, this 16-bit is used to identify the relative byte position of the data in this fragment with respect to the start of the data submitted to CLNP; i.e., it indicates where in the original datagram this fragment belongs. IETF Page 9 August 10, 1992 CLNP for TUBA Internet Draft 4.15 Options All CLNP options are of the form <parameter code>, <parameter lenth>, and <parameter value>. Both the parameter code and length fields are always one byte long; the length parameter value, in bytes, is indicated in the parameter length field. The following options are defined for CLNP for TUBA. [Editor's note: whether the CLNP Priority option may be useful in TUBA envirnoments is for further study. I have omitted this option in this draft.] 4.15.1 _S_e_c_u_r_i_t_y The value of the parameter code field is binary 1100 0101. The length field must be set to the length of a basic or extended security option as identified in RFC1108 [10], plus 1. Byte 1 of the security parameter value field is set to a binary value 0100 0000, indicating that the remaining bytes of the security field contain either the basic or extended security options as identified in RFC 1108 [10]. The processing of the security option is defined in RFC1108. This encoding points to the administration of the source address (e.g., ISOC) as the administration of the security option; it is distinguished from the globally unique format whose definition is reserved for OSI. 4.15.2 _T_y_p_e__o_f__S_e_r_v_i_c_e The value of the parameter code field must be set to a value of binary 1100 0011. The length field must be set to the length of the type of service field as identified in RFCxxxx [11], plus 1. Byte 1 of the type of service parameter field is set to a binary value 0100 0000, indicating that the remaining bytes of the type of service field are to be encoded as described in RFCxxxx. The processing of the type of service option is defined in RFCxxxx. 4.15.3 _P_a_d_d_i_n_g The padding field is used to lengthen the packet header to a convenient size. The parameter code field must be set to a value of binary 1100 1100. The value of the parameter length field is variable. The parameter value may contain any value. 4.15.4 _S_o_u_r_c_e__R_o_u_t_i_n_g The Complete and Partial Source Route options of CLNP replace the strict and loose source route options of IP, respectively. The parameter code for Source Routing is binary 1100 1000. The length of the source routing parameter value is variable. The first byte of the parameter value is a type code, indicating IETF Page 10 Internet Draft CLNP for TUBA August 10, 1992 Complete Source Routing (binary 0000 0001) or partial source routing (binary 0000 0000). A complete description of the encoding of the parameter values of this option is deferred until the format and semantics of addressing is better defined. 4.15.5 _R_e_c_o_r_d__R_o_u_t_e The Record route option of CLNP replaces the record route option. The parameter code for Record Route is binary 1100 1011. The length of the record route parameter value is variable. A complete description of the parameter value of this option is deferred until format and semantics of addressing is better defined. 4.15.6 _T_i_m_e_s_t_a_m_p There is no timestamp option in CLNP. Two choices exist: define the option and submit it to ISO, or steal a parameter code point from the many that are reserved. This paper proposes that the parameter code value 1110 1110 be used to identify the timestamp option, and that the syntax and semantics of Timestamp be identical to that defined in IP. The issue of whether the code point is usurped or codified in ISO 8473 is left open. The Timestamp is defined in RFC 791. It is proposed that the parameter code 1110 1110 be used rather than the option type code 68 to identify the Timestamp option, and that the parameter value convey the option length. Byte 1 of the Timestamp parameter value shall be encoded as the pointer (byte 3 of IP Timestamp); byte 2 of the parameter value shall be encoded as the overflow/format byte (byte 4 of IP Timestamp); the remaining bytes shall be used to encode the timestamp list. The size is fixed by the source, and cannot be changed to accommodate additional timestamp information. 5. REFERENCES IETF Page 11 August 10, 1992 CLNP for TUBA Internet Draft [1] ISO 8473--International Standards Organization--Data Communications-- Protocol for Providing the Connectionless-mode Network Service [2] Callon, R., TCP/UDP over Bigger Addresses (TUBA), Request for Comments 1347, Network Information Center, SRI International, Menlo Park, CA, May 1992. [3] Postel, J., Transmission ControlProtocol. Request for Comments 793, Network Information Center, SRI International, Menlo Park, CA, 1981 September. [4] RFC768, User Datagram Protocol. Request for Comments 768, Network Information Center, SRI International, Menlo Park, CA <date> [5] Postel, J., Internet Protocol. Request for Comments 791, Network Information Center, SRI International, Menlo Park, CA, 1981 September. [6] RFC994, ISO CLNP, Draft International Standard version. [7] ISO8348--International Standards Organization--Data Communications--OSI Network Layer Addressing [8] RFCiiii, Addressing for the new Internet [9] RFCjjjj, Assigned Numbers [10] RFC1108, Kent, S., Security Option for IP. [11] RFCxxxx, Almquist, P., Type of Service Appendix A. Checksum Algorithms (from ISO 8473) Symbols used in algorithms: c0, c1 variables used in the algorithms i position of byte in header (first byte is i=1) Bi value of byte i in the header n position of first byte of checksum (n=8) L Length of header in bytes X Value of byte one of the checksum parameter Y Value of byte two of the checksum parameter IETF Page 12 Internet Draft CLNP for TUBA August 10, 1992 Addition is performed in on of the two following modes: o+ module 255 arithmetic; o+ eight-bit one's complement arithmetic; The algorithm for Generating the Checksum Parameter Value is as follows: A. Construct the complete header with the value of the checksum parameter field set to zero; i.e., c0 <- c1 <- 0; B. Process each byte of the header sequentially from i=1 to L by: o+ c0 <- c0 + Bi o+ c1 <- c1 + c0 C. Calculate X, Y as follows: o+ X <- (L - 8)(c0 - c1) modulo 255 o+ Y <- (L - 7)(-C0) + c1 D. If X = 0, then X <- 255 E. If Y = 0, then Y <- 255 F. place the values of X and Y in bytes 8 and 9 of the header, respectively The algorithm for checking the value of the checksum parameter is as follows: A. If bytes 8 and 9 of the header both contain zero, then the checksum calculation has succeeded; else if either but not both of these bytes contains the value zero then the checksum is incorrect; otherwise, initialize: c0 <- c1 <- 0 B. Process each octet of the header sequentially from i = 1 to L by: o+ c0 <- c0 + Bi o+ c1 <- c1 + c0 C. When all the bytes have been processed, if c0 = c1 = 0, then the checksum calculation has succeeded, else it has failed. There is a separate algorithm to adjust the checksum parameter value when a byte has been modified (such as the TTL). Suppose IETF Page 13 August 10, 1992 CLNP for TUBA Internet Draft the value in byte k is changed by Z = newvalue - oldvalue. If X and Y denote the checksum values held in bytes n and n+1 respectively, then adjust X and Y as follows: If X = 0 and Y = 0 then do nothing, else if X = 0 or Y = 0 then the checksum is incorrect, else: X <- (k - n - 1)Z + X modulo 255 Y <- (n - k)Z + Y modulo 255 If X = 0, then X <- 255; if Y = 0, then Y <- 255. In the example, n = 89; if the byte altered is the TTL (byte 4), then k = 4. For the case where the lifetime is decreased by one unit (Z = -1), the assignment statements for the new values of X and Y in the immediately preceeding algorithm simplify to: X <- X + 5 Modulo 255 Y <- Y - 4 Modulo 255
- CLNP for TUBA Dave Piscitello
- CLNP for TUBA Dave Piscitello
- Re: CLNP for TUBA Beast (Donald E. Eastlake, 3rd)
- CLNP for TUBA minshall