Network Working Group Eric C. Rosen Internet Draft Yakov Rekhter Expiration Date: September 1997 Daniel Tappan Dino Farinacci Guy Fedorkow Cisco Systems, Inc. March 1997 Label Switching: Label Stack Encodings draft-rosen-tag-stack-01.txt Status of this Memo This document is an Internet-Draft. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet-Drafts. Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress." To learn the current status of any Internet-Draft, please check the "1id-abstracts.txt" listing contained in the Internet-Drafts Shadow Directories on ftp.is.co.za (Africa), nic.nordu.net (Europe), munnari.oz.au (Pacific Rim), ds.internic.net (US East Coast), or ftp.isi.edu (US West Coast). Abstract "Label Switching" [1] requires a set of procedures for augmenting network layer packets with "Label Stacks" (formerly called "Tag Stacks"), thereby turning them into "Labeled packets". This document specifies the encoding to be used, on PPP data links and LAN data links, in order to produce a Labeled packet from a Label Stack and a network layer packet. This document also specifies rules and procedures for processing the various fields of the Label Stack encoding. Rosen, et al. [Page 1] =0C Internet Draft draft-rosen-tag-stack-01.txt March 1997 Table of Contents 1 Introduction ........................................... 2 1.1 Specification of Requirements .......................... 3 2 The Label Stack ........................................ 3 2.1 Encoding the Label Stack ............................... 3 2.2 Determining the Network Layer Protocol ................. 6 2.3 Processing the Time to Live Field ...................... 6 2.3.1 Definitions ............................................ 6 2.3.2 Protocol-independent rules ............................. 6 2.3.3 IP-dependent rules ..................................... 7 3 Fragmentation and Path MTU Discovery ................... 7 3.1 Terminology ............................................ 8 3.2 Maximum Initially Labeled IP Datagram Size ............. 9 3.3 When are Labeled IP Datagrams Too Big? ................. 10 3.4 Processing Labeled IP Datagrams which are Too Big ...... 11 3.5 Implications with respect to Path MTU Discovery ........ 12 3.5.1 Tunneling through a Transit Routing Domain ............. 12 3.5.2 Tunneling Private Addresses through a Public Backbone .. 13 4 Transporting Labeled Packets over PPP .................. 13 4.1 Introduction ........................................... 13 4.2 A PPP Network Control Protocol for Label Switching ..... 14 4.3 Sending Labeled Packets ................................ 15 4.4 Label Switching Control Protocol Configuration Options . 15 5 Transporting Labeled Packets over LAN Media ............ 16 6 Security Considerations ................................ 16 7 Authors' Addresses ..................................... 16 8 References ............................................. 17 1. Introduction [1] describes a set of procedures for augmenting network layer packets with "Label Stacks" (formerly called "Tag Stacks"), thereby turning them into "Labeled packets". This document specifies the encoding to be used, on PPP data links and LAN data links, in order to produce a Labeled packet from a Label Stack and a network layer packet. This document also specifies rules and procedures for processing the various fields of the Label Stack encoding. Label Switching itself is independent of any particular network layer protocol; however, while most of the relevant procedures are independent of the network Rosen, et al. [Page 2] =0C Internet Draft draft-rosen-tag-stack-01.txt March 1997 layer protocol, some procedures differ for different protocols. In this document, we specify the protocol-independent procedures, but we specify protocol-dependent procedures only for IPv4. 1.1. Specification of Requirements In this document, several words are used to signify the requirements of the specification. These words are often capitalized. MUST This word, or the adjective "required", means that the definition is an absolute requirement of the specification. MUST NOT This phrase means that the definition is an absolute prohibition of the specification. SHOULD This word, or the adjective "recommended", means that there may exist valid reasons in particular circumstances to ignore this item, but the full implications must be understood and carefully weighed before choosing a different course. MAY This word, or the adjective "optional", means that this item is one of an allowed set of alternatives. An implementation which does not include this option MUST be prepared to interoperate with another implementation which does include the option. 2. The Label Stack 2.1. Encoding the Label Stack On both PPP and LAN data links, the Label Stack is represented as a sequence of Label Stack Entries. Each Label Stack Entry is represented by 4 octets. This is shown in Figure 1. Rosen, et al. [Page 3] =0C Internet Draft draft-rosen-tag-stack-01.txt March 1997 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 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Labe= l | Label |rsvd |CoS|S| TTL | Stac= k +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Entr= y Label: Label Value, 19 bits rsvd: Reserved, 3 bits CoS: Class of Service, 2 bits S: Bottom of Stack, 1 bit TTL: Time to Live, 7 bits Figure 1 The Label Stack Entries appear AFTER the data link layer headers, but BEFORE any network layer headers. The top of the Label Stack appears earliest in the packet, and the bottom appears latest. The network layer packet immediately follows the Label Stack Entry which has the S bit set. Each Label Stack Entry is broken down into the following fields: 1. Bottom of Stack (S) This bit is set to one for the last entry in the Label Stack (i.e., for the bottom of the stack), and zero for all other Label Stack Entries. 2. Time to Live (TTL) This seven-bit field is used to encode a time-to-live value. The processing of this field is described in section 2.3. 3. Class of Service (CoS) This two-bit field is used to identify a "Class of Service". Presumably the setting of this field will affect the scheduling and/or discard algorithms which are applied to the packet as it is transmitted through the network. When an unlabeled packet is initially labeled, the value assigned to the CoS field in the Label Stack Entry is determined by policy. Some possible policies are: Rosen, et al. [Page 4] =0C Internet Draft draft-rosen-tag-stack-01.txt March 1997 - the CoS value is a function of the IP ToS value - the CoS value is a function of the packet's input interface - the CoS value is a function of the "flow type" Many other policies are also possible. When an additional Label is pushed onto the Stack of a packet that is already labeled: - in general, the value of the CoS field in the new top stack entry should be equal to the value of the CoS field of the old top stack entry; - however, in some cases, most likely at boundaries between network service providers, the value of the CoS field in the new top stack entry may be determined by policy. 4. Reserved These three bits are reserved. They MUST be set to zero when writing, and MUST be ignored when reading. 5. Label Value This 19-bit field carries the actual value of the Label. When a Labeled packet is received, the Label value at the top of the Stack is looked up. As a result of this lookup one learns: (a) all the information needed to forward the packet (b) the operation to be performed on the Label Stack before forwarding; this operation may be to replace the top Label Stack Entry with another, or to pop Entries off the Label Stack, or to push Entries on the Label Stack, or any combination of these operations. There are several reserved Label values: i. A value of 0 represents the "IPv4 Explicit NULL Label". This Label value is only legal when it is the sole Label Stack Entry. It indicates that the Label Stack must be popped, and the forwarding of the packet must then be based on the IP header. Rosen, et al. [Page 5] =0C Internet Draft draft-rosen-tag-stack-01.txt March 1997 ii. A value of 1 represents the "Router Alert Label". This Label value is legal anywhere in the Label Stack except at the bottom. When a received packet contains this Label value at the top of the Label Stack, it is delivered to a local software module for processing. The actual forwarding of the packet is determined by the Label beneath it in the stack. However, before the packet is forwarded, the Router Alert Label should be pushed back onto its Label Stack. 2.2. Determining the Network Layer Protocol When the last Label is popped from the Label Stack, it is necessary to determine the particular network layer protocol which is being carried. Since the Label header carries no explicit field to identify the network layer header, this must be inferable from the value of the Label which is popped. 2.3. Processing the Time to Live Field 2.3.1. Definitions The "incoming TTL" of a Labeled packet is defined to be the value of the TTL field in the Label Stack Entry which is at the top of the Label Stack when the packet is received. The "outgoing TTL" of a Labeled packet is defined to be the larger of: (a) one less than the incoming TTL, (b) zero. 2.3.2. Protocol-independent rules If the outgoing TTL of a Labeled packet is 0, then the Labeled packet MUST NOT be further forwarded; the packet's lifetime in the network is considered to have expired. Depending on the Label value in the Label Stack Entry, the packet MAY be silently discarded, or the packet MAY have its Label Stack stripped off, and passed as an unlabeled packet to the ordinary processing for network layer packets which have exceeded their maximum lifetime in the network. However, even if the Label Stack is stripped, the packet MUST NOT be further forwarded. Rosen, et al. [Page 6] =0C Internet Draft draft-rosen-tag-stack-01.txt March 1997 When a Labeled packet is forwarded, the TTL field of the Label Stack Entry at the top of the Label Stack must be set to the outgoing TTL value. Note that the outgoing TTL value is a function solely of the incoming TTL value, and is independent of whether any Labels are pushed or popped before forwarding. 2.3.3. IP-dependent rules When an IP packet is first Labeled, the TTL field of the Label Stack Entry is set to the smaller of 127 and the value of the IP TTL field. When a Label is popped, and the resulting Label Stack is empty, then: (a) if the value in the IP TTL field is less than or equal to 127, it MUST be replaced with the outgoing TTL value, as defined above; (b) if the value in the IP TTL field is greater than 127, then the new value of the IP TTL field MUST be set to: (Old_IP_TTL_value - 127 + Outgoing_TTL) 3. Fragmentation and Path MTU Discovery Just as it is possible to receive an unlabeled IP datagram which is too large to be transmitted on its output link, it is possible to receive a Labeled packet which is too large to be transmitted on its output link. It is also possible that a received packet (Labeled or unlabeled) which was originally small enough to be transmitted on that link becomes too large by virtue of having one or more additional Labels pushed onto its Label Stack. In Label switching, a packet may grow in size if additional Labels get pushed on. Thus if one receives a Labeled packet with a 1500-byte frame payload, and pushes on an additional Label, one needs to forward it as frame with a 1504-byte payload. This section specifies the rules for processing Labeled packets which are "too large". In particular, it provides rules which ensure that hosts implementing RFC 1191 Path MTU Discovery will be able to generate IP datagrams that do not need fragmentation, even if they get Labeled as the traverse the network. Rosen, et al. [Page 7] =0C Internet Draft draft-rosen-tag-stack-01.txt March 1997 In general, hosts which do not implement RFC 1191 Path MTU Discovery send IP datagrams which contain no more than 576 bytes; the probability that such datagrams will need to get fragmented, even if they get labeled, is very small, since the MTUs in use on most data links today are 1500 bytes or greater. Some hosts that do not implement RFC 1191 Path MTU Discovery will generate IP datagrams containing 1500 bytes, as long as the IP Source and Destination addresses are on the same subnet. These datagrams will not pass through routers, and hence will not get fragmented. Unfortunately, some hosts will generate IP datagrams containing 1500 bytes, as long the IP Source and Destination addresses do not have the same classful network number. This is the one case in which there is significant risk of fragmentation when such datagrams get labeled. This document specifies procedures which allow one to configure the network so that large datagrams from hosts which do not implement Path MTU Discovery get fragmented just once, when they are first labeled. These procedures make it possible (assuming suitable configuration) to avoid any need to fragment packets which have already been Labeled. 3.1. Terminology With respect to a particular data link, we can use the following terms: - Frame Payload: The contents of a data link frame, excluding any data link layer headers or trailers (e.g., MAC headers, LLC headers, 802.1q or 802.1p headers, PPP header, frame check sequences, etc.). When a frame is carrying an an unlabeled IP datagram, the Frame Payload is just the IP datagram itself. When a frame is carrying a Labeled IP datagram, the Frame Payload consists of the Label Header and the IP datagram. - Conventional Maximum Frame Payload Size: The maximum Frame Payload size allowed by standards. For example, the Conventional Maximum Frame Payload Size for ethernet is 1500 bytes. Rosen, et al. [Page 8] =0C Internet Draft draft-rosen-tag-stack-01.txt March 1997 - True Maximum Frame Payload Size: The maximum size frame payload which can be sent and received properly by the interface hardware attached to the data link. On ethernet and 802.3 networks, it is believed that the True Maximum Frame Payload Size is 4-8 bytes larger than the Conventional Maximum Frame Payload Size (unless an 802.1q or 802.1p header is present). For example, it is believed that most ethernet equipment could correctly send and receive packets carrying a payload of 1504 or perhaps even 1508 bytes, at least, as long as the ethernet header does not have an 802.1q or 802.1p field. On PPP links, the True Maximum Frame Payload Size may be virtually unbounded. - Effective Maximum Frame Payload Size for Labeled Packets: This is either be the Conventional Maximum Frame Payload Size or the True Maximum Frame Payload Size, depending on the capabilities of the equipment on the data link and the size of the ethernet header being used. - Initially Labeled IP Datagram Suppose that an unlabeled IP datagram is received at a particular Label Switching Router (LSR), and that the the LSR pushes on a Label before forwarding the datagram. Such a datagram will be called an Initially Labeled IP Datagram at that LSR. - Previously Labeled IP Datagram An IP datagram which had already been Labeled before it was received by a particular LSR. 3.2. Maximum Initially Labeled IP Datagram Size Every Label Switching Router which is capable of (a) receiving an unlabeled IP datagram, (b) adding a Label Stack to the datagram, and Rosen, et al. [Page 9] =0C Internet Draft draft-rosen-tag-stack-01.txt March 1997 (c) forwarding the resulting Labeled packet, MUST support a configuration parameter known as the "Maximum IP Datagram Size for Labeling", which may be set to a non-negative value. If this configuration parameter is set to zero, it has no effect. If it is set to a positive value, it is used in the following way. If: (a) an unlabeled IP datagram is received, and (b) that datagram does not have the DF bit set in its IP header, and (c) that datagram needs to be labeled before being forwarded, and (d) the size of the datagram (before labeling) exceeds the value of the parameter, then (a) the datagram must be broken into fragments, each of whose size is no greater than the value of the parameter, and (b) each fragment must be labeled and then forwarded. If this configuration parameter is set to a value of 1488, for example, then any unlabeled IP datagram containing more than 1488 bytes will be fragmented before being labeled. Each fragment will be capable of being carried on a 1500-byte data link, without further fragmentation, even if as many as three Labels are pushed onto its Label Stack. In other words, setting this parameter to a non-zero value allows one to eliminate all fragmentation of Previously Labeled IP Datagrams, but it may cause some unnecessary fragmentation of Initially Labeled IP Datagrams. Note that the parameter has no effect on IP Datagrams that have the DF bit set, which means that it has no effect on Path MTU Discovery. 3.3. When are Labeled IP Datagrams Too Big? A Labeled IP datagram whose size exceeds the Conventional Maximum Frame Payload Size of the data link over which it is to be forwarded MUST be considered to be "too big". A Labeled IP datagram whose size exceeds the True Maximum Frame Payload Size of the data link over which it is to be forwarded MAY be considered to be "too big". A Labeled IP datagram which is not "too big" MUST be transmitted Rosen, et al. [Page 10] =0C Internet Draft draft-rosen-tag-stack-01.txt March 1997 without fragmentation. 3.4. Processing Labeled IP Datagrams which are Too Big If a Labeled IP datagram is "too big", and the DF bit is not set in its IP header, then the Label Switching Router MAY discard the datagram. Note that discarding such datagrams is a sensible procedure only if the "Maximum Initially Labeled IP Datagram Size" is set to a non-zero value in every Label Switching Router in the network which is capable of adding a Label Stack to an unlabeled IP datagram. If the Label Switching Router chooses not to discard a Labeled IP datagram which is too big, or if the DF bit is set in that datagram, then it MUST execute the following algorithm: 1. Strip off the Label header to obtain the IP datagram. 2. Let N be the number of bytes in the Label Stack (i.e, 4 times the number of Label Stack Entries). 3. If the IP datagram does NOT have the "Don't Fragment" bit set in its IP header: a. convert it into fragments, each of which MUST be at least N bytes less than the Effective Maximum Frame Payload Size. b. Prepend each fragment with the same Label header that would have been on the original datagram had fragmentation not been necessary. c. Forward the fragments 4. If the IP datagram has the "Don't Fragment" bit set in its IP header: a. the datagram MUST NOT be forwarded b. Create an ICMP Destination Unreachable Message: i. set its Code field (RFC 792) to "Fragmentation Required and DF Set", Rosen, et al. [Page 11] =0C Internet Draft draft-rosen-tag-stack-01.txt March 1997 ii. set its Next-Hop MTU field (RFC 1191) to the difference between the Effective Maximum Frame Payload Size and the value of N c. If possible, transmit the ICMP Destination Unreachable Message to the source of the of the discarded datagram. 3.5. Implications with respect to Path MTU Discovery The procedures described above for handling datagrams which have the DF bit set, but which are "too large", have an impact on the Path MTU Discovery procedures of RFC 1191. Hosts which implement these procedures will discover an MTU which is small enough to allow n Labels to be pushed on the datagrams, without need for fragmentation, where n is the number of Labels that actually get pushed on along the path currently in use. In other words, datagrams from hosts that use Path MTU Discovery will never need to be fragmented due to the need to put on a Label header, or to add new Labels to an existing Label header. (Also, datagrams from hosts that use Path MTU Discovery generally have the DF bit set, and so will never get fragmented anyway.) However, note that Path MTU Discovery will only work properly if, at the point where a Labeled IP Datagram's fragmentation needs to occur, it is possible to route to the packet's source address. If this is not possible, then the ICMP Destination Unreachable message cannot be sent to the source. 3.5.1. Tunneling through a Transit Routing Domain Suppose one is using Label switching to "tunnel" through a transit routing domain, where the external routes are not leaked into the domain's interior routers. If a packet needs fragmentation at some router within the domain, and the packet's source address is an external address, and the packet's DF bit is set, it is desirable to be able to originate an ICMP message at that router and have it routed correctly to the source of the fragmented packet. However, that source is an external address, which is not known to the internal routers. Therefore, in order for Path MTU Discovery to work, in any routing domain in which external routes are not leaked into the interior routers, there MUST be a default route which causes all packets carrying external destination addresses to be sent to a border router. Rosen, et al. [Page 12] =0C Internet Draft draft-rosen-tag-stack-01.txt March 1997 For example, one of the border routers may inject "default" into the IGP. 3.5.2. Tunneling Private Addresses through a Public Backbone In other cases where Label switching is used to tunnel through a routing domain, it may not be possible to route to the source address of a fragmented packet at all. This would be the case, for example, if the IP addresses carried in the packet were private addresses, and Label Switching were being used to tunnel those packets through a public backbone. In such cases, the Label Switching Router at the transmitting end of the tunnel MUST be able to determine the MTU of the tunnel as a whole. It SHOULD do this by sending packets through the tunnel to the tunnel's receiving endpoint, and performing Path MTU Discovery with those packets. Then any time the transmitting endpoint of the tunnel needs to send a packet into the tunnel, and that packet has the DF bit set, and it exceeds the tunnel MTU, the transmitting endpoint of the tunnel MUST send the ICMP Destination Unreachable message to the source, with code "Fragmentation Required and DF set", and the Next-Hop MTU Field set as described above. 4. Transporting Labeled Packets over PPP The Point-to-Point Protocol (PPP) [PPP] provides a standard method for transporting multi-protocol datagrams over point-to-point links. PPP defines an extensible Link Control Protocol, and proposes a family of Network Control Protocols for establishing and configuring different network-layer protocols. This section defines the Network Control Protocol for establishing and configuring Label Switching over PPP. 4.1. Introduction PPP has three main components: 1. A method for encapsulating multi-protocol datagrams. 2. A Link Control Protocol (LCP) for establishing, configuring, and testing the data-link connection. Rosen, et al. [Page 13] =0C Internet Draft draft-rosen-tag-stack-01.txt March 1997 3. A family of Network Control Protocols for establishing and configuring different network-layer protocols. In order to establish communications over a point-to-point link, each end of the PPP link must first send LCP packets to configure and test the data link. After the link has been established and optional facilities have been negotiated as needed by the LCP, PPP must send Label Switching Control packets to enable the transmission of Labeled packets. Once the Label Switching Control Protocol has reached the Opened state, Labeled packets can be sent over the link. The link will remain configured for communications until explicit LCP or Label Switching Control Protocol packets close the link down, or until some external event occurs (an inactivity timer expires or network administrator intervention). 4.2. A PPP Network Control Protocol for Label Switching The Label Switching Control Protocol (LSCP) is responsible for enabling and disabling the use of Label switching on a PPP link. it uses the same packet exchange mechanism as the Link Control Protocol (LCP). LSCP packets may not be exchanged until PPP has reached the Network-Layer Protocol phase. LSCP packets received before this phase is reached should be silently discarded. The Label Switching Control Protocol is exactly the same as the Link Control Protocol [1] with the following exceptions: 1. Frame Modifications The packet may utilize any modifications to the basic frame format which have been negotiated during the Link Establishment phase. 2. Data Link Layer Protocol Field Exactly one LSCP packet is encapsulated in the PPP Information field, where the PPP Protocol field indicates type hex 80?? (Label Switching). 3. Code field Only Codes 1 through 7 (Configure-Request, Configure-Ack, Configure-Nak, Configure-Reject, Terminate-Request, Terminate- Ack and Code-Reject) are used. Other Codes should be treated as unrecognized and should result in Code-Rejects. Rosen, et al. [Page 14] =0C Internet Draft draft-rosen-tag-stack-01.txt March 1997 4. Timeouts LSCP packets may not be exchanged until PPP has reached the Network-Layer Protocol phase. An implementation should be prepared to wait for Authentication and Link Quality Determination to finish before timing out waiting for a Configure-Ack or other response. It is suggested that an implementation give up only after user intervention or a configurable amount of time. 5. Configuration Option Types None. 4.3. Sending Labeled Packets Before any Labeled packets may be communicated, PPP must reach the Network-Layer Protocol phase, and the Label Switching Control Protocol must reach the Opened state. Exactly one Labeled packet is encapsulated in the PPP Information field, where the PPP Protocol field indicates either type hex 00?? (Label Switching -- Unicast) or type hex 00?? (Label Switching -- Multicast). The maximum length of a Labeled packet transmitted over a PPP link is the same as the maximum length of the Information field of a PPP encapsulated packet. The format of the Information field itself is as defined in section 2. Note that two codepoints are defined for Labeled packets; one for multicast and one for unicast. Once the LSCP has reached the Opened state, both Label Switched multicasts and Label Switched unicasts can be sent over the PPP link. 4.4. Label Switching Control Protocol Configuration Options There are no configuration options. Rosen, et al. [Page 15] =0C Internet Draft draft-rosen-tag-stack-01.txt March 1997 5. Transporting Labeled Packets over LAN Media A pair of two byte ethertype values will be obtained, one representing "Label Switching -- Unicast" and one representing "Label Switching -- Multicast". These can be used with either the Ethernet encapsulation or the 802.3 SNAP/SAP encapsulation to carry Labeled packets. Exactly one Labeled packet is carried in each frame. The Label Stack Entries immediately precede the network layer header, and follow any data link layer headers, including any VLAN headers that may exist. 6. Security Considerations Security considerations are not discussed in this document. 7. Authors' Addresses Eric C. Rosen Cisco Systems, Inc. 250 Apollo Drive Chelmsford, MA, 01= 824 E-mail: erosen@cisco.com Dan Tappan Cisco Systems, Inc. 250 Apollo Drive Chelmsford, MA, 01824 E-mail: tappan@cisco.com Dino Farinacci Cisco Systems, Inc. 170 Tasman Drive San Jose, CA, 95134 E-mail: dino@cisco.com Yakov Rekhter Cisco Systems, Inc. 170 Tasman Drive San Jose, CA, 95134 E-mail: yakov@cisco.com Rosen, et al. [Page 16] =0C Internet Draft draft-rosen-tag-stack-01.txt March 1997 Guy Fedorkow Cisco Systems, Inc. 250 Apollo Drive Chelmsford, MA, 01824 E-mail: fedorkow@cisco.com 8. References [1] "Tag Switching Architecture - Overview", 1/9/97, draft-rekhter- tagswitch-arch-00.txt, Rekhter, Davie, Katz, Rosen, Swallow [2] "Internet Control Message Protocol", RFC 792, 9/81, Postel [3] "Path MTU Discovery", RFC 1191, 11/90, Mogul & Deering Rosen, et al. [Page 17]