100 likes | 215 Views
Group-to-RP Mapping. IETF Draft: draft-ietf-pim-group-rp-mapping-03 Authors: Andy Kessler, Cisco Systems Bharat Joshi, Infosys Technologies Ltd. David McWalter, Metaswitch Networks. Background. Outgrowth of issues found in standardizing new PIM MIB 2 years of discussion and revisions
E N D
Group-to-RP Mapping IETF Draft: draft-ietf-pim-group-rp-mapping-03 Authors: Andy Kessler, Cisco Systems Bharat Joshi, Infosys Technologies Ltd. David McWalter, Metaswitch Networks
Background • Outgrowth of issues found in standardizing new PIM MIB • 2 years of discussion and revisions • Draft now contains considerable input and ideas
Need for Standard Algorithm Need for standard deterministic algorithm • No existing RFC/Draft defines a complete algorithm • A standard algorithm is required for interop • Resolves cases in which there are several group mappings for a single group from different sources • A standard algorithm could be used to determine the selected RP by an independent network management system
Use Cases • Default static Group-to-RP mappings with dynamically learned entries • Default infrastructure for general purpose apps and specific mappings for special apps that change more often • Migration situations • After acquisitions or mergers companies will need deterministic method to move from one RP mapping method to another (e.g. AutoRP to BSR) • Use by Management Systems • Ability for NMS to determine Group-RP-Mappings used by a router
What is new in this Rev ? New in this Rev • Removed Override Dynamic flag • Filtering mappings at the edge is recommended • RP Priority is now used for BSR learned mappings Fixed in previous Version • Hash function is applied for BSR learned PIM-SM mappings • BSR and AutoRP are not recommended to run together Algorithm is now consistent with several implementations
Proposed Algorithm The RP would be selected with this preference: • Embedded RP • Longest match • PIM-Bidir over PIM-SM • Origin preference - BSR, AutoRP, static and other • RP Priority – if mapping was learned through BSR • Hash Function – if BSR and PIM-SM • Highest IP address
Current Issues with Draft • None ? • Ready for WGLC ?
References – PIM-SM RFC • Section 4.7.1 of RFC 4601 - PIM-SM Protocol Spec • Describes the existing algorithm for choosing a group from multiple Group-to-RP Mapping. It does not address these issues • Does not consider the origin of a group-to-rp mapping (e.g. bsr, autorp, static, etc) • Doesn't allow for higher priority of PIM-Bidir over PIM-SM • Section 4.7.2 of RFC 4601 - PIM-SM Protocol Spec • Describes the Hash function and talks about the highest IP address but only after the hash function.
References – BSR RFC • Section 1.1 of RFC 5059 - BSR Mechanism for PIM • ... the router chooses only one of the RPs by applying a deterministic algorithm so that all routers in the domain make the same choice. It is important to note that this algorithm is part of the specification of the individual routing protocols (and may differ among them), not of the BSR specification. • Section 3.3 of RFC 5059 - BSR Mechanism for PIM • If the BSR learns of both BIDIR and PIM-SM Candidate-RPs for the same group range, the BSR MUST only include RPs for one of the protocols in the BSMs. The default behavior SHOULD be to prefer BIDIR.
References – Embedded RP RFC • Section 7.1 of RFC 3956 - Embedded RP for IPv6 • To avoid loops and inconsistencies, for addresses in the range FF70::/12, the Embedded-RP mapping MUST be considered the longest possible match and higher priority than any other mechanism.