--- 1/draft-ietf-mboned-embeddedrp-00.txt 2006-02-05 00:19:23.000000000 +0100 +++ 2/draft-ietf-mboned-embeddedrp-01.txt 2006-02-05 00:19:23.000000000 +0100 @@ -1,22 +1,22 @@ mboned Working Group P. Savola Internet Draft CSC/FUNET -Expiration Date: April 2004 +Expiration Date: August 2004 B. Haberman Caspian Networks - October 2003 + February 2004 - Embedding the Address of RP in IPv6 Multicast Address +Embedding the Rendezvous Point (RP) Address in an IPv6 Multicast Address - draft-ietf-mboned-embeddedrp-00.txt + draft-ietf-mboned-embeddedrp-01.txt Status of this Memo This document is an Internet-Draft and is subject to all provisions of Section 10 of RFC2026. 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. @@ -27,535 +27,581 @@ material or to cite them other than as "work in progress." The list of current Internet-Drafts can be accessed at http://www.ietf.org/ietf/1id-abstracts.txt. To view the list Internet-Draft Shadow Directories, see http://www.ietf.org/shadow.html. Abstract - There exists a huge deployment problem with global, interdomain IPv6 - multicast: Protocol Independent Multicast - Sparse Mode (PIM-SM) - Rendezvous Points (RPs) have no way of communicating the information - about multicast sources to other multicast domains, as there is no - Multicast Source Discovery Protocol (MSDP), and the whole interdomain - Any Source Multicast (ASM) model is rendered unusable; Source - Specific Multicast (SSM) avoids these problems but is not considered - readily deployable at the moment. This memo defines a PIM-SM group- - to-RP mapping which encodes the address of the RP in the IPv6 - multicast address. In consequence, there would be no need for - interdomain MSDP, and even intra-domain RP configuration could be - simplified. This memo updates RFC 3306. + A very difficult deployment problem with global, interdomain IPv6 + multicast using Protocol Independent Multicast - Sparse Mode (PIM-SM) + has been identified. This memo defines an address allocation policy + in which the address of the Rendezvous Point (RP) is encoded in the + IPv6 multicast group address. For PIM-SM, this can be seen as a + specification of a group-to-RP mapping mechanism. This allows an + easy deployment of scalable inter-domain multicast, and simplifies + the intra-domain multicast configuration as well. This memo updates + the addressing format presented in RFC 3306. Table of Contents 1. Introduction ............................................... 2 2. Unicast-Prefix-based Address Format ........................ 4 3. Modified Unicast-Prefix-based Address Format ............... 4 4. Embedding the Address of the RP in the Multicast Address ... 5 5. Examples ................................................... 6 5.1. Example 1 .............................................. 6 - 5.2. Example 2 .............................................. 6 - 5.3. Example 3 .............................................. 6 + 5.2. Example 2 .............................................. 7 + 5.3. Example 3 .............................................. 7 5.4. Example 4 .............................................. 7 - 6. Operational Requirements ................................... 7 - 6.1. Anycast-RP ............................................. 7 - 6.2. Guidelines for Assigning IPv6 Addresses to RPs ......... 7 - 7. Required PIM-SM Modifications .............................. 7 - 7.1. Overview of the Model .................................. 9 - 8. Scalability/Usability Analysis ............................. 9 + 6. Operational Considerations ................................. 7 + 6.1. RP Redundancy .......................................... 7 + 6.2. RP Deployment .......................................... 8 + 6.3. Guidelines for Assigning IPv6 Addresses to RPs ......... 8 + 7. PIM-SM Protocol Modifications .............................. 8 + 7.1. PIM-SM Group-to-RP Mapping ............................. 9 + 7.2. Overview of the Model .................................. 9 + 8. Scalability/Usability Analysis ............................. 10 9. Acknowledgements ........................................... 11 10. Security Considerations ................................... 11 - 11. References ................................................ 12 - 11.1. Normative References .................................. 12 - 11.2. Informative References ................................ 12 - Authors' Addresses ............................................. 13 - A. Discussion about Design Tradeoffs .......................... 13 - Intellectual Property Statement ................................ 14 - Full Copyright Statement ....................................... 15 + 11. References ................................................ 13 + 11.1. Normative References .................................. 13 + 11.2. Informative References ................................ 13 + Authors' Addresses ............................................. 14 + A. Discussion about Design Tradeoffs .......................... 14 + B. Changes since -00 .......................................... 15 + Intellectual Property Statement ................................ 15 + Full Copyright Statement ....................................... 16 1. Introduction - As has been noticed [V6MISSUES], there exists a huge deployment - problem with global, interdomain IPv6 multicast: PIM-SM [PIM-SM] RPs - have no way of communicating the information about multicast sources - to other multicast domains, as there is no MSDP [MSDP], and the whole - interdomain Any Source Multicast model is rendered unusable; SSM - [SSM] avoids these problems. + As has been noticed [V6MISSUES], there exists a deployment problem + with global, interdomain IPv6 multicast: PIM-SM [PIM-SM] RPs have no + way of communicating the information about multicast sources to other + multicast domains, as there is no Multicast Source Discovery Protocol + (MSDP) [MSDP] (at least yet). Therefore the whole interdomain Any + Source Multicast model is rendered unusable; Source-Specific + Multicast (SSM) [SSM] avoids these problems but is not a complete + solution for several reasons. - It has been noted that there are some problems with SSM deployment - and support: it seems unlikely that SSM could be usable as the only - interdomain multicast routing mechanism in the short term. This memo - proposes a fix to interdomain multicast routing, and provides an - additional method for the RP discovery with the intra-domain case. + Further, it has been noted that there are some problems with the + support and deployment of mechanisms SSM would require: it seems + unlikely that SSM could be usable as the only interdomain multicast + routing mechanism in the short term. - This document proposes a solution to the group-to-RP mapping problem - which leverages and extends [RFC3306] by encoding the RP address of - the IPv6 multicast group into the group address itself. + This memo describes a multicast address allocation policy in which + the address of the RP is encoded in the IPv6 multicast group address, + and specifies a PIM-SM group-to-RP mapping to use the encoding, + leveraging and extending the unicast-prefix -based addressing + [RFC3306]. This mechanism not only provides a simple solution for IPv6 - interdomain ASM but can be used as a simple solution for IPv6 - intradomain ASM on scoped addresses, as well. The use as a substitute - for Bootstrap Router protocol (BSR) [BSR] is also possible. + interdomain Any Source Multicast (ASM) but can be used as a simple + solution for IPv6 intradomain ASM on scoped addresses as well. It + can also be used in those deployment scenarios which would have + previously used the Bootstrap Router protocol (BSR) [BSR]. - The solution consists of two elements applicable to a subrange of - [RFC3306] IPv6 multicast group addresses which are defined by setting - one previously unused bit of the Flags field to "1": + The solution consists of three elements: + + o A specification of a subrange of [RFC3306] IPv6 multicast group + addresses defined by setting one previously unused bit of the + Flags field to "1", o A specification of the mapping by which such a group address encodes the RP address that is to be used with this group, and o A specification of optional and mandatory procedures to operate ASM with PIM-SM on these IPv6 multicast groups. - Addresses in this subrange will be called embedded-RP addresses. If - used in the interdomain, a mechanism similar to MSDP is not required - for these addresses and RP configuration for these addresses can be - as simple as zero configuration for routers supporting this - specification. + Addresses in the subrange will be called embedded RP addresses. This + scheme obviates the need for inter-domain MSDP, and the routers are + not required to include any multicast configuration, except when they + act as an RP. - It is self-evident that a 128 bit RP address can in general not be - embedded into a 128-bit group address with space left to carry a - group identity itself. An appropriate form of encoding is thus - defined, and it is assumed that the Interface-ID of RPs in the - embedded-RP range can be assigned to be specific values. + In general, a 128-bit RP address can't be embedded into a 128-bit + group address with space left to carry the group identity itself. An + appropriate form of encoding is thus defined, and it is assumed that + the Interface-ID of RPs in the embedded RP range can be assigned to + be a specific value. If these assumptions can't be followed, either operational procedures and configuration must be slightly changed or this mechanism can not be used. The assignment of multicast addresses is outside the scope of this - document; however, the mechanisms are very probably similar to ones - used with [RFC3306]. + document; it is up to the RP and applications to ensure that group + addresses are unique using some unspecified method. However, the + mechanisms are very probably similar to ones used with [RFC3306]. + + Similarly, RP failure management methods, such as Anycast-RP, are out + of scope for this document. These do not work without additional + specification or deployment. This is covered briefly in Section 6.1. This memo updates the addressing format presented in RFC 3306. The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in [RFC2119]. 2. Unicast-Prefix-based Address Format As described in [RFC3306], the multicast address format is as follows: | 8 | 4 | 4 | 8 | 8 | 64 | 32 | +--------+----+----+--------+--------+----------------+----------+ |11111111|flgs|scop|reserved| plen | network prefix | group ID | +--------+----+----+--------+--------+----------------+----------+ - Where flgs are "0011". (The first two bits are yet undefined and - thus zero.) + Where flgs are "0011". (The first two bits are yet undefined, sent + as zero and ignored on receipt.) 3. Modified Unicast-Prefix-based Address Format - This memo proposes a modification to the unicast-prefix-based address - format: + This memo specifies a modification to the unicast-prefix-based + address format: 1. If the second high-order bit in "flgs" is set to 1, the address of the RP is embedded in the multicast address, as described in this memo. - 2. If the second high-order bit in "flgs" was set to 1, interpret + 2. If the second high-order bit in "flgs" is set to 1, interpret the last low-order 4 bits of "reserved" field as signifying the RP interface ID, as described in this memo. In consequence, the address format becomes: | 8 | 4 | 4 | 4 | 4 | 8 | 64 | 32 | +--------+----+----+----+----+--------+----------------+----------+ - |11111111|flgs|scop|rsvd|RPad| plen | network prefix | group ID | + |11111111|flgs|scop|rsvd|RIID| plen | network prefix | group ID | +--------+----+----+----+----+--------+----------------+----------+ +-+-+-+-+ flgs is a set of 4 flags: |0|R|P|T| +-+-+-+-+ - R = 1 indicates a multicast address that embeds the address of the - PIM-SM RP. Then P MUST BE set to 1, and consequently T MUST be set - to 1, as specified in [RFC3306]. + R = 1 indicates a multicast address that embeds the address on the + RP. Then P MUST BE set to 1, and consequently T MUST be set to 1, as + specified in [RFC3306]. - In the case that R = 1, the last 4 bits of previously reserved field - ("RPad") are interpreted as embedding the interface ID of the RP, as + In the case that R = 1, the last 4 bits of the previously reserved + field are interpreted as embedding the RP interface ID ("RIID"), as specified in this memo. R = 0 indicates a multicast address that does not embed the address - of the PIM-SM RP and follows the semantics defined in [ADDRARCH] and - [RFC3306]. In this context, the value of "RPad" has no meaning. + of the RP and follows the semantics defined in [ADDRARCH] and + [RFC3306]. In this context, the value of "RIID" MUST be as zero and + MUST be ignored on receipt. 4. Embedding the Address of the RP in the Multicast Address The address of the RP can only be embedded in unicast-prefix -based ASM addresses. To identify whether an address is a multicast address as specified in this memo and to be processed any further, it must satisfy all of the below: o it MUST be a multicast address and have R, P, and T flag bits set - to 1 (that is, be part of the prefix FF7::/12 or FFF::/12), + to 1 (that is, be part of the prefixes FF70::/12 or FFF0::/12), o "plen" MUST NOT be 0 (ie. not SSM), and o "plen" MUST NOT be greater than 64. The address of the RP can be obtained from a multicast address - satisfying the above criteria by taking the following steps: - - 1. take the last 96 bits of the multicast address add 32 zero bits - at the end, + satisfying the above criteria by taking the two steps: - 2. zero the last 128-"plen" bits, and + 1. copy the first "plen" bits of the "network prefix" to a zeroed + 128-bit address structure, and + 2. replace the last 4 bits with the contents of "RIID". - 3. replace the last 4 bits with the contents of "RPad". + These two steps could be illustrated as follows: - One should note that there are several operational scenarios when - [RFC3306] statement "all non-significant bits of the network prefix - field SHOULD be zero" is ignored -- and why the second step, above, - is necessary. This is to allow multicast address assignments to - third parties which still use your RP; see example 2 below. + | 20 bits | 4 | 8 | 64 | 32 | + +---------+----+----+----------------+----------+ + |xtra bits|RIID|plen| network prefix | group ID | + +---------+----+----+----------------+----------+ + || \\ vvvvvvvvvvv + || ``====> copy plen bits of "network prefix" + || +------------+------------------------+ + || | network pre| 0000000000000000000000 | + || +------------+------------------------+ + \\ + ``=================> copy RIID to the last 4 bits + +------------+---------------------+--+ + | network pre| 0000000000000000000 |ID| + +------------+---------------------+--+ + One should note that there are several operational scenarios (see + Example 2 below) when [RFC3306] statement "all non-significant bits + of the network prefix field SHOULD be zero" is ignored. This is to + allow multicast address assignments to third parties which still use + the RP associated with the network prefix. "plen" higher than 64 MUST NOT be used as that would overlap with the upper bits of multicast group-id. - The implementation MUST perform at least the same address validity - checks to the calculated RP address as to one received via other - means (like BSR [BSR] or MSDP for IPv4), to avoid e.g. the address - being "::" or "::1". + When processing an encoding to get the RP address, the multicast + routers MUST perform at least the same address validity checks to the + calculated RP address as to one received via other means (like BSR + [BSR] or MSDP for IPv4), to avoid e.g., the address being "::", + "::1", or a link-local address. - One should note that the 4 bits reserved for "RPad" set the upper - bound for RPs per multicast group address; not the number of RPs in a - subnet, PIM-SM domain or large-scale network. + One should note that the 4 bits reserved for "RIID" set the upper + bound for RPs for the combination of scope, network prefix, and group + ID -- without varying any of these, you can have 4 bits worth of + different RPs. However, each of these is an IPv6 group address of + its own (i.e., there can be only one RP per multicast address). 5. Examples + Four examples of multicast address allocation and resulting group-to- + RP mappings are described here, to better illustrate the + possibilities provided by the encoding. + 5.1. Example 1 - The network administrator of 3FFE:FFFF::/32 wants to set up an RP for - the network and all of his customers. He chooses network - prefix=3FFE:FFFF and plen=32, and wants to use this addressing - mechanism. The multicast addresses he will be able to use are of the - form: + The network administrator of 2001:DB8::/32 wants to set up an RP for + the network and all of his customers. (S)he chooses network + prefix=2001:DB8 and plen=32, and wants to use this addressing + mechanism. The multicast addresses (s)he will be able to use are of + the form: - FF7x:y20:3FFE:FFFF:zzzz:zzzz: + FF7x:y20:2001:DB8:zzzz:zzzz: Where "x" is the multicast scope, "y" the interface ID of the RP address, and "zzzz:zzzz" will be freely assignable within the PIM-SM domain. In this case, the address of the PIM-SM RP would be: - 3FFE:FFFF::y + 2001:DB8::y - (and "y" could be anything from 0 to F); the address 3FFE:FFFF::y/128 + (and "y" could be anything from 0 to F); the address 2001:DB8::y/128 is added as a Loopback address and injected to the routing system. 5.2. Example 2 - As above, the network administrator can also allocate multicast - addresses like "FF7x:y20:3FFE:FFFF:DEAD::/80" to some of his + As in Example 1, the network administrator can also allocate + multicast addresses like "FF7x:y20:2001:DB8:DEAD::/80" to some of his customers within the PIM-SM domain. In this case the RP address - would still be "3FFE:FFFF::y". + would still be "2001:DB8::y". Note the second rule of deriving the RP address: the "plen" field in - the multicast address, (hex)20 = 32, refers to the length of "network + the multicast address, 0x20 = 32, refers to the length of "network prefix" field considered when obtaining the RP address. In this - case, only the first 32 bits of the network prefix field, "3FFE:FFFF" + case, only the first 32 bits of the network prefix field, "2001:DB8" are preserved: the value of "plen" takes no stance on actual unicast/multicast prefix lengths allocated or used in the networks, - here from 3FFE:FFFF:DEAD::/48. + here from 2001:DB8:DEAD::/48. 5.3. Example 3 - In the above network, the network admin sets up addresses as above, - but an organization wants to have their own PIM-SM domain; that's - reasonable. The organization can pick multicast addresses like - "FF7x:y30:3FFE:FFFF:BEEF::/80", and then their RP address would be - "3FFE:FFFF:BEEF::y". + In the network of Examples 1 and 2, the network admin sets up + addresses for use by their customers, but an organization wants to + have their own PIM-SM domain; that's reasonable. The organization + can pick multicast addresses like "FF7x:y30:2001:DB8:BEEF::/80", and + then their RP address would be "2001:DB8:BEEF::y". 5.4. Example 4 In the above networks, if the admin wants to specify the RP to be in - a non-zero /64 subnet, he could always use something like - "FF7x:y40:3FFE:FFFF:BEEF:FEED::/96", and then their RP address would - be "3FFE:FFFF:BEEF:FEED::y". There are still 32 bits of multicast + a non-zero /64 subnet, (s)he could always use something like + "FF7x:y40:2001:DB8:BEEF:FEED::/96", and then their RP address would + be "2001:DB8:BEEF:FEED::y". There are still 32 bits of multicast group-id's to assign to customers and self. -6. Operational Requirements +6. Operational Considerations -6.1. Anycast-RP + This desction describes the major operational considerations for + those deploying this mechanism. - One should note that MSDP is also used, in addition to interdomain - connections between RPs, in anycast-RP [ANYCASTRP] -technique, for - sharing the state information between different RPs in one PIM-SM - domain. However, there are other propositions, like [ANYPIMRP]. +6.1. RP Redundancy - Anycast-RP mechanism is incompatible with this addressing method - unless MSDP is specified and implemented. Alternatively, another - method for sharing state information could be used. + A technique called "Anycast RP" is used within a PIM-SM domain to + share an address and multicast state information between a set of + RP's mainly for redundancy purposes. Typically, MSDP has been used + for that as well [ANYCASTRP]. There are also other approaches, like + using PIM for sharing this information [ANYPIMRP]. - Anycast-RP and other possible RP failover mechanisms are outside of - the scope of this memo. + RP failover cannot be used with this specification without additional + mechanisms or techniques such as MSDP, PIM-SM extensions, or + anycasting the RP address in the IGP without state sharing (depending + on the redundancy requirements, this may or may not be enough, + though). However, the redundancy mechanisms are outside of the scope + of this memo. -6.2. Guidelines for Assigning IPv6 Addresses to RPs +6.2. RP Deployment + + As there is no need to share inter-domain state with MSDP, each DR + connecting multicast sources could act as an RP without scalability + concerns about MSDP sessions. + + This might be particularly attractive when concerned about RP + redundancy. In the case where the DR close to a major source for a + group acts as the RP, a certain amount of fate-sharing properties can + be obtained without using any RP failover mechanisms: if the DR goes + down, the multicast transmission may not be all that interesting + anymore in any case. + + Along the same lines, it's may also be desirable to distribute the RP + responsibilities to multiple RPs. As long as different RPs serve + different groups, this is is trivial: each group should map to a + different RP (or enough many different RPs that the load on one RP is + not a problem). However, load sharing one group faces the similar + challenges as Anycast-RP. + +6.3. Guidelines for Assigning IPv6 Addresses to RPs With this mechanism, the RP can be given basically any network prefix up to /64. The interface identifier will have to be manually - configured to match "RPad". + configured to match "RIID". - RPad = 0 SHOULD NOT be used as using it would cause ambiguity with + RIID = 0 SHOULD NOT be used as using it would cause ambiguity with the Subnet-Router Anycast Address [ADDRARCH]. If an administrator wishes to use an RP address that does not conform to the addressing topology but is still from the network provider's - prefix (e.g. an additional loopback address assigned on a router), + prefix (e.g., an additional loopback address assigned on a router), that address can be injected into the routing system via a host route. -7. Required PIM-SM Modifications - - The use of multicast addresses with embedded RP addresses requires - additional PIM-SM processing. Namely, a PIM-SM router will need to - be able to recognize the encoding and derive the RP address from the - address using the rules in section 4 and to be able to use the - embedded RP, instead of its own for multicast addresses in this - specified range. +7. PIM-SM Protocol Modifications - The three key places where these modifications are used are the - Designated Routers (DRs) on the receiver/sender networks, the - backbone networks, and the RPs in the domain where the embdedded - address has been derived from (see figure below). + This section describes how PIM-SM is modified, i.e., how the group- + to-RP mapping mechanism works for Embedded RP. - For the foreign DRs (rtrR1, rtrR23, and rtrR4), this means sending - PIM-SM Join/Prune/Register messages towards the foreign RP (rtrRP_S). - Naturally, PIM-SM Register-Stop and other messages must also be - allowed from the foreign RP. DRs in the local PIM-SM domain (rtrS) - do the same. +7.1. PIM-SM Group-to-RP Mapping - For the RP (rtrRP_S), this means being able to recognize and validate - PIM-SM messages which use RP-embedded addressing originated from any - DR at all. + The only PIM-SM modification required is implementing this mechanism + as one group-to-RP mapping method. - For the other routers on the path (rtrBB), this means recognizing and - validating that the Join/Prune PIM-SM messages using the embedded RP - addressing are on the right path towards the RP they think is in - charge of the particular address. + The implementation will have to recognize the address format and + derive and use the RP address using the rules in Section 4. This + information is used at least when performing RPF lookups and when + processing Join/Prune messages, or performing Register-encapsulation. - nodeS - rtrS - rtrRP_S - rtrBB -----+--- rtrR1 - node1 - | | | - node2_S ---------+ | +-- rtrR23 - node2 - | | - | +---- node3 - | - +------------ rtrR4 - node4 + To avoid loops and inconsistancies, the group-to-RP mapping specified + in this memo MUST be used for all embedded RP groups (i.e., with + prefix FF70::/12 or FFF0::/12). - In addition, the administration of the PIM-SM domains MAY have an - option to manually override the RP selection for the embedded RP - multicast addresses: the default policy SHOULD be to use the embedded - RP. + It is worth noting that compared to the other group-to-RP mappings, + which can be precomputed, the embedded RP mapping must be redone for + every new IPv6 group address which would map to a different RP. For + efficiency, the results may be cached in an implementation-specific + manner. - The extraction of the RP information from the multicast address - should be done during forwarding state creation. That is, if no - state exists for the multicast address, PIM-SM must take the embedded - RP information into account when creating forwarding state. Unless - otherwise dictated by the administrative policy, this would result in - a receiver's DR initiating a PIM-SM Join towards the foreign RP or a - source's DR sending PIM-SM Register messages towards the foreign RP. + This group-to-RP mapping mechanism must be supported by the DR + adjacent to senders and any router on the path from any receiver to + the RP. It also must be supported by any router on the path from any + sender to the RP -- in case the RP issues a Register-Stop and Joins + the sources. It should be noted that this approach removes the need to run inter- domain MSDP. Multicast distribution trees in foreign networks can be joined by issuing a PIM-SM Join/Prune/Register to the RP address encoded in the multicast address. Also, the addressing model described here could be used to replace or augment the intra-domain Bootstrap Router mechanism (BSR), as the RP- - mappings can be communicated by the multicast address assignment. + mappings can be derived from the application of multicast address + assignmen policies. -7.1. Overview of the Model +7.2. Overview of the Model + + This section gives a high level, non-normative overview of how + Embedded RP operates, as specified in the previous section. The steps when a receiver wishes to join a group are: - 1. A receiver finds out a group address from some means (e.g. SDR + 1. A receiver finds out a group address from some means (e.g., SDR or a web page). + 2. The receiver issues an MLD Report, joining the group. 3. The receiver's DR will initiate the PIM-SM Join process towards the RP embedded in the multicast address. The steps when a sender wishes to send to a group are: 1. A sender finds out a group address from some means, whether in - an existing group (e.g. SDR, web page) or in a new group (e.g. - a call to the administrator for group assignment, use of a - multicast address assignment protocol). + an existing group (e.g., SDR, web page) or in a new group + (e.g., a call to the administrator for group assignment, use of + a multicast address assignment protocol). 2. The sender sends to the group. 3. The sender's DR will send the packets unicast-encapsulated in PIM-SM Register-messages to the RP address encoded in the multicast address (in the special case that DR is the RP, such sending is only conceptual). - In both cases, the messages then go on as specified in [PIM-SM] and - other specifications (e.g. Register-Stop and/or SPT Join); there is - no difference in them except for the fact that the RP address is - derived from the multicast address. - - Sometimes, some information, using conventional mechanisms, about - another RP exists in the PIM-SM domain. The embedded RP SHOULD be - used by default, but there MAY be an option to switch the preference. - This is because especially when performing PIM-SM forwarding in the - transit networks, the routers must have the same notion of the RP, or - else the messages may be dropped. + In fact, all the messages go as specified in [PIM-SM] -- embedded RP + just acts as a group-to-RP mapping mechanism; instead of obtaining + the address of the RP from local configuration or configuration + protocols (e.g., BSR), it is derived transparently from the encoded + multicast address. 8. Scalability/Usability Analysis Interdomain MSDP model for connecting PIM-SM domains is mostly - hierarchical. The "embedded RP address" changes this to a mostly - flat, sender-centered, full-mesh virtual topology. - - This may or may not cause some effects; it may or may not be - desirable. At the very least, it makes many things much more robust - as the number of third parties is minimized. A good scalability - analysis is needed. - - In some cases (especially if e.g. every home user is employing site- - local multicast), some degree of hierarchy would be highly desirable, - for scalability (e.g. to take the advantage of shared multicast - state) and administrative point-of-view. - - Being able to join/send to remote RPs has security considerations - that are considered below, but it has an advantage too: every group - has a "home RP" which is able to control (to some extent) who are - able to send to the group. + hierarchical in configuration and deployment, but flat with regard to + information distribution. The embedded RP inter-domain model behaves + as if all of the Internet was a single PIM domain, with just one RP + per group. So, the inter-domain multicast becomes a flat, RP- + centered topology. The scaling issues are be described below. - One should note that the model presented here simplifies the PIM-SM - multicast routing model slightly by removing the RP for senders and - receivers in foreign domains. One scalability consideration should - be noted: previously foreign sources sent the unicast-encapsulated - data to their local RP, now they do so to the foreign RP responsible - for the specific group. This is especially important with large - multicast groups where there are a lot of heavy senders -- - particularly if implementations do not handle unicast-decapsulation - well. + Previously foreign sources sent the unicast-encapsulated data to + their local RP, now they do so to the foreign RP responsible for the + specific group. This is especially important with large multicast + groups where there are a lot of heavy senders -- particularly if + implementations do not handle unicast-decapsulation well. This model increases the amount of Internet-wide multicast state slightly: the backbone routers might end up with (*, G) and (S, G, - rpt) state between receivers and the RP, in addition to (S, G) states - between the receivers and senders. Certainly, the amount of inter- - domain multicast traffic between sources and the embedded-RP will - increase compared to the ASM model with MSDP; however, the domain - responsible for the RP is expected to be able to handle this. + rpt) state between receivers (and past receivers, for PIM Prunes) and + the RP, in addition to (S, G) states between the receivers and + senders. Certainly, the amount of inter-domain multicast traffic + between sources and the embedded RP will increase compared to the ASM + model with MSDP. - As the address of the RP is tied to the multicast address, in the - case of RP failure PIM-SM BSR mechanisms cannot pick a new RP; the - failover mechanisms, if used, for backup RPs are different, and - typically would depend on sharing one address. The failover - techniques are outside of the scope of this memo. + The embedded RP model is practically identical in both inter-domain + and intra-domain cases to the traditional PIM-SM in intra-domain. On + the other hand, PIM-SM has been deployed (in IPv4) in inter-domain + using MSDP; compared to that inter-domain model, this specification + simplifies the multicast routing by removing the RP for senders and + receivers in foreign domains. + + As the address of the RP is tied to the multicast address, the RP + failure management becomes more difficult, as failover or redundancy + mechanisms (e.g., BSR, Anycast-RP with MSDP) cannot be used as-is. + This described briefly in Section 6.1. The PIM-SM specification states, "Any RP address configured or - learned MUST be a domain-wide reachable address". What this means is - not clear, even without embedded-RP. However, typically this - statement cannot be proven especially with the foreign RPs (typically - one can not even guarantee that the RP exists!). The bottom line is - that while traditionally the configuration of RPs and DRs was - typically a manual process, and e.g. configuring a non-existant RP - was possible, but here the hosts and users which use multicast - indirectly specify the RP. + learned MUST be a domain-wide reachable address". What "reachable" + precisely means is not clear, even without embedded RP. This + statement cannot be proven especially with the foreign RPs (one can + not even guarantee that the RP exists!). Instead of configuring RPs + and DRs with a manual process (configuring a non-existent RP was + possible though rare), with this specification the hosts and users + using multicast indirectly specify the RP themselves, lowering the + expectancy of the RP reachability. + + Being able to join/send to remote RPs raises security concerns that + are considered separately, but it has an advantage too: every group + has a "home RP" which is able to control (to some extent) who are + able to send to the group. + + A more extensive description and comparison of the inter-domain + multicast routing models (traditional ASM with MSDP, embedded RP, + SSM) and their security properties has been described in [PIMSEC]. 9. Acknowledgements Jerome Durand commented on an early draft of this memo. Marshall Eubanks noted an issue regarding short plen values. Tom Pusateri noted problems with earlier SPT-join approach. Rami Lehtonen pointed out issues with the scope of SA-state and provided extensive - commentary. Nidhi Bhaskar gave the draft a thorough review. The - whole MboneD working group is also acknowledged for the continued - support and comments. + commentary. Nidhi Bhaskar gave the draft a thorough review. + Toerless Eckert, Hugh Holbrook, and Dave Meyer provided very + extensive feedback. The whole MboneD working group is also + acknowledged for the continued support and comments. 10. Security Considerations - The address of the PIM-SM RP is embedded in the multicast address. - RPs may be a good target for Denial of Service attacks -- as they are - a single point of failure (excluding failover techniques) for a - group. In this way, the target would be clearly visible. However, it - could be argued that if interdomain multicast was to be made work - e.g. with MSDP, the address would have to be visible anyway (through - via other channels, which may be more easily securable). + The address of the RP is encoded in the multicast address. RPs may + be a good target for Denial of Service attacks -- as they are a + single point of failure (excluding failover techniques) for a group. + In this way, the target would be clearly visible. However, it could + be argued that if interdomain multicast was to be made to work e.g., + with MSDP, the address would have to be visible anyway (through via + other channels). As any RP will have to accept PIM-SM Join/Prune/Register messages from any DR, this might cause a potential DoS attack scenario. However, this can be mitigated by the fact that the RP can discard - all such messages for all multicast addresses that do not embed the + all such messages for all multicast addresses that do not encode the address of the RP, and if deemed important, the implementation could also allow manual configuration of which multicast addresses or prefixes embedding the RP could be used, so that only the pre-agreed sources could use the RP. In a similar fashion, when a receiver joins to an RP, the DRs must - accept similar PIM-SM messages back RPs. + accept similar PIM-SM messages back from RPs. - One consequence of the usage model is that it allows Internet-wide - multicast state creation (from receiver(s) in another domain to the - RP in another domain) compared to the domain wide state creation in - the MSDP model. + One consequence of the embedded RP usage model is that it allows + Internet-wide multicast state creation (from receiver(s) in another + domain to the RP in another domain) compared to the domain wide state + creation in the traditional ASM model. However, the traditional ASM + model also requires MSDP state to propagate everywhere in inter- + domain, so the total amount of state is smaller. One should observe that the embedded RP threat model is actually pretty similar to SSM; both mechanisms significantly reduce the threats at the sender side, but have new ones in the receiver side, - as any receiver can try to join any non-existant group or channel, - and the local DR or RP cannot readily reject such joins (based on - MSDP information). + as any receiver can try to join any non-existent group or channel, + and the local DR or RP cannot readily reject (e.g., based on MSDP + information) such joins. - RPs may become a bit more single points of failure as anycast-RP - mechanism is not (at least immediately) available. This can be - partially mitigated by the fact that some other forms of failover are - still possible, and there should be less need to store state as with - MSDP. + RPs become single points of failure as anycast-RP mechanism is not + (at least immediately) available. However, some other forms of + failover are still possible (see Section 6.1) and one can obtain some + forms of fate-sharing properties with a proper placement of RPs (see + Section 6.2). The implementation MUST perform at least the same address validity checks to the embedded RP address as to one received via other means - (like BSR or MSDP), to avoid the address being e.g. "::" or "::1". + (like BSR or MSDP), to avoid the address being e.g., "::", "::1", or + a link-local address. + + A more extensive description and comparison of the inter-domain + multicast routing models (traditional ASM with MSDP, embedded RP, + SSM) and their security properties has been described in [PIMSEC]. 11. References 11.1. Normative References [ADDRARCH] Hinden, R., Deering, S., "IP Version 6 Addressing Architecture", RFC3513, April 2003. [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997. [RFC3306] Haberman, B., Thaler, D., "Unicast-Prefix-based IPv6 Multicast Addresses", RFC3306, August 2002. 11.2. Informative References [ANYCASTRP] Kim, D. et al, "Anycast RP mechanism using PIM and MSDP", RFC 3446, January 2003. [ANYPIMRP] Farinacci, D., Cai, Y., "Anycast-RP using PIM", - work-in-progress, draft-farinacci-pim-anycast-rp-00.txt, - January 2003. + work-in-progress, draft-ietf-pim-anycast-rp-00.txt, + November 2003. [BSR] Fenner, B., et al., "Bootstrap Router (BSR) Mechanism for PIM Sparse Mode", work-in-progress, draft-ietf-pim-sm- bsr-03.txt, February 2003. - [MSDP] Meyer, D., Fenner, B, (Eds.), "Multicast Sourc - Discovery Protocol (MSDP)", work-in-progress, - draft-ietf-msdp-spec-20.txt, May 2003. + [MSDP] Meyer, D., Fenner, B, (Eds.), "Multicast Source + Discovery Protocol (MSDP)", RFC 3618, October 2003. + + [PIMSEC] Savola, P., Lehtonen, R., Meyer, D., "PIM-SM Multicast + Routing Security Issues and Enhancements", + work-in-progress, draft-savola-mboned-mroutesec-00.txt, + January 2004. [PIM-SM] Fenner, B. et al, "Protocol Independent Multicast - Sparse Mode (PIM-SM): Protocol Specification (Revised), work-in-progress, draft-ietf-pim-sm-v2-new-08.txt, October 2003. [SSM] Holbrook, H. et al, "Source-Specific Multicast for IP", - work-in-progress, draft-ietf-ssm-arch-03.txt, - May 2003. + work-in-progress, draft-ietf-ssm-arch-04.txt, + October 2003. [V6MISSUES] Savola, P., "IPv6 Multicast Deployment Issues", work-in-progress, draft-savola-v6ops-multicast- issues-02.txt, October 2003. Authors' Addresses Pekka Savola CSC/FUNET Espoo, Finland @@ -563,82 +609,72 @@ Brian Haberman Caspian Networks One Park Drive, Suite 300 Research Triangle Park, NC 27709 EMail: brian@innovationslab.net Phone: +1-919-949-4828 A. Discussion about Design Tradeoffs - The initial thought was to use only SPT join from local RP/DR to - foreign RP, rather than a full PIM Join to foreign RP. However, this - turned out to be problematic, as this kind of SPT joins where - disregarded because the path had not been set up before sending them. - A full join to foreign PIM domain is a much clearer approach. - - One could argue that there should be more RPs than the 4-bit "RPad" + One could argue that there should be more RPs than the 4-bit "RIID" allows for, especially if anycast-RP cannot be used. In that light, - extending "RPad" to take full advantage of whole 8 bits would seem + extending "RIID" to take full advantage of whole 8 bits would seem reasonable. However, this would use up all of the reserved bits, and leave no room for future flexibility. In case of large number of RPs, an operational workaround could be to split the PIM domain: for example, using two /33's instead of one /32 would gain another 16 (or 15, if zero is not used) RP addresses. Note that the limit of 4 bits worth of RPs just depends on the prefix the RP address is derived from; one can use multiple prefixes in a domain, and the limit of 16 (or 15) RPs should never really be a problem. - Some hierarchy (e.g. two-level, "ISP/customer") for RPs could - possibly be added if necessary, but that would be torturing one 128 - bits even more. - - One particular case, whether in the backbone or in the sender's - domain, is where the regular PIM-SM RP would be X, and the embedded - RP address would be Y. This could e.g. be a result of a default all- - multicast-to-one-RP group mapping, or a local policy decision. The - embedded RP SHOULD be used by default, but there MAY be an option to - change this preference. - Values 64 < "plen" < 96 would overlap with upper bits of the multicast group-id; due to this restriction, "plen" must not exceed 64 bits. This is in line with RFC 3306. The embedded RP addressing could be used to convey other information (other than RP address) as well, for example, what should be the RPT threshold for PIM-SM. These could be encoded in the RP address - somehow, or in the multicast group address. However, such - modifications are beyond the scope of this memo. + somehow, or in the multicast group address. Whether this is a good + idea is another thing. In any case, such modifications are beyond + the scope of this memo. - Some kind of rate-limiting functions, ICMP message responses, or - similar could be defined for the case of when the RP embedded in the - address is not willing to serve for the specific group (or doesn't - even exist). Typically this would result in the datagrams getting - blackholed or rejected with ICMP. In particular, a case for - "rejection" or "source quench" -like messages would be in the case - that a source keeps transmitting a huge amount of data, which is sent - to a foreign RP using Register message but is discarded if the RP - doesn't allow the source host to transmit: the RP should be able to - indicate to the DR, "please limit the amount of Register messages", - or "this source sending to my group is bogus". Note that such "kiss- - of-death" packets have an authentication problem; spoofing them could - result in an entirely different kind of Denial of Service, for - legitimate sources. One possibility here would be to specify some - form of "return routability" check for DRs and RPs; for example, if a - DR receives packets from a host to group G G (resulting in RP address - R), the DR would send only a limited amount of packets to R until it - has heard back from R (a "positive acknowledgement"). It is not - clear whether this needs to be considered or specified in more - detail. + For the cases where the RPs do not exist or are unreachable, or too + much state is being generated to reach in a resource exhaustion DoS + attack, some forms of rate-limiting or other mechanisms could be + deployed to mitigate the threats while trying not to disturb the + legitimate usage. This has been described at more length in + [PIMSEC]. - Could this model work with bidir-PIM? Is it feasible? Not sure, not - familiar enough with bidir-PIM. + The mechanism is not usable with Bidirectional PIM without protocol + extensions, as pre-computing the Designated Forwarder is not + possible. + +B. Changes since -00 + + [[ RFC-Editor: please remove before publication ]] + + o Lots of editorial cleanups, or cleanups without techinical + changes. + o Reinforce the notion of Embedded RP just being a group-to-RP + mapping mechanism (causing substantive rewriting in section 7); + highlight the fact that precomputing the group-to-RP mapping is + not possible. + o Add (a bit) more text on RP redundancy and deployment tradeoffs + wrt. RPs becoming SPoF. + o Clarify the usability/scalability issues in section 8. + o Clarify the security issues in Sections 8, Security + Considerations and Appendix A, mainly by referring to a separate + document. + o Add a MUST that embedded RP mappings must be honored by + implementations. Intellectual Property Statement The IETF takes no position regarding the validity or scope of any intellectual property or other rights that might be claimed to pertain to the implementation or use of the technology described in this document or the extent to which any license under such rights might or might not be available; neither does it represent that it has made any effort to identify any such rights. Information on the IETF's procedures with respect to rights in standards-track and