IP MULTICAST

1. IP MULTICAST • IP Address => Host Identifier THE ABILITY TO DELIVER AN IP PACKET( DATAGRAM ) TO MULTIPLE DESTINATION ( HOST GROUP )

2. IP MULTICAST SUPPORT • Multicast IP Address • Mapping to LAN • Group Management Protocol at IP Level • HOST GROUP Multicast Router Host A Host B Host C

3. IP MULTICAST ADDRESS 11100000000000000000000000000000 11101111111111111111111111111111 224.0.0.0 239.255.255.255 • IP Address Space: Class D • Well-known IP Multicast Address 1110 224.0.0.0 224.0.0.1 224.0.0.2 224.0.0.3 224.0.0.4 224.0.0.5 224.0.0.6 224.0.0.7 224.0.0.8 224.0.0.9 224.0.0.10 224.0.0.11 224.0.0.12 - 224.0.0.255 Base Address (Reserved) All Systems on this Subnet All Routers on this Subnet Unassigned DVMRP Routers OSPFIGP OSPFIGP All Router OSPFIGP OSPFIGP Designated Routers ST Routers STHosts RIP2 Routers IGRP Routers Mobile-Agents Unassigned

4. IP MULTICAST ADDRESS TO ETHERNET MAPPING • Ethernet: 48 bit -> 23 bit 23 bit 1110 Class D Address Group ID copied to Ethernet address 0000000100000000010111100 Ethetnet Address 0 1 0 0 5 E

5. HOST GROUP • Set of Hosts Listening to A Particular IP Multicast Address • A Set of Zero or More Hosts • Membership -> Dynamic • Hosts는 언제나 가입/탈퇴 • Member의 수와 위치 제한 없음 • Permanent or Transient • All Hosts Group • 224.0.0.1 • All systems on this subnet • Automatically join

6. ACTIVATION multicast router multicast router multicast router 2. Start delay timer for group membership ( 0 ~ D seconds) 3. A hosts expires the delay timer 1. Query to all hosts group multicast router multicast router 4. The host sends Report 5. Hosts in the group stop the delay timer

7. IGMP MESSAGE FORMAT 0 3 4 7 8 15 16 31 Version Type Unused Checksum Group Address IP header IGMP message 8 bytes 20 bytes

8. MULTICAST ROUTING PROTOCOL IGMP Multicast Routing Protocol: MOSPF, DVMRP,..

9. MULTICAST ROUTING ALGORITHMS • Flooding • Spanning Tree • Reverse-Path Forwarding (RPF) • Reverse-Path Forwarding with Prunes • Steiner Tree • Core-Based Tree

10. FLOODING • Simplest Multicast Routing Algorithm • Procedure • receive a multicast packet • test for the first reception • if first, forward the packet on all (exception the incoming interface) interfaces • Insufficient Use of Router Memory

11. SPANNING TREE • A Tree Structure Where Only One Active Path Connects Any Routers (No Loop) Router Leaf

12. SPANNING TREE (CONTINUED) • Multicast Packets are Forwarded Along the Spanning Tree • One Spanning Tree for the Entire Internet • Easy to Implement • Traffic is Centralized on a Small Number of Links • Group Membership not Considered • Does not Provide the Most Efficient Path Between Members

13. 1 2 D A E D C B B E C A E D C A B 3 4 5 6 REVERSE PATH FORWARDING (RPF) • Different Spanning Tree for Each Active (Source, Group) Pair • When a multicast packet is received, note source (S) and Interface (I) • If I belongs to the shortest path toward S, forward to all interfaces except I. • If the test in the above is false, refuse the packet. 1 2 1 2 3 3 4 4 5 5 6 6 From Source C From Source A

14. RPF (Continued) • Guarantee the Fastest Possible Delivery, as Multicasting Follows the Shortest Path from Source to Destination • Because a Different Tree is Computed for Each Source, the Packet Are Spread over Multiple Links Resulting in better Network Utilization • Group Membership is not Considered

15. RPF WITH PRUNES • Group Membership Considered • Avoid Forwarding Datagrams onto a Subnet Without Members • The first packet is flooded on the whole network • Prune messages are sent back from leaf nodes without group members • Propagated back to source, forming minimal tree • Drawbacks • The first packet is flooded on the whole network • The router must keep states per group and source • To Accommodate Dynamic Membership Changes, Prune Information is Removed from the Memory of all Routers at Regular Intervals. The Next Packet is Forwarded all Leaf Nodes followed by a Series of Prune Messages • Not Scalable

16. C E A E D C B A D B STEINER TREE • Minimize the Number of Links to Connect the Group Members • Source: C, Reception: A, D 1 2 1 2 3 4 3 4 5 5 6 6 Steiner Tree RPF Tree

17. CORE BASED TREE (CBT) • Core - a Fixed Point is the Center of the Multicast Group • Recipients Send JOIN Message to the Core, to form a Tree • One Spanning Tree per Group • Traffic Concentration Problem • Scalable • Variation - Multiple Cores

18. CBT • A single bidirectional shared tree for a group • Adv. • reduction of state info. That needs to be maintained at each router • Disadv. • traffic concentration • delivery delay can be higher than in source-based shortest path trees

19. DVMRP Distance vector multicast routing protocol Reverse-path distance vector (I.e., router calculates reverse optimum path, from source to itself) MOSPF Multicast open shortest path first Uses group membership packet Creates a shortest-path spanning tree Multicast Routing

20. DVMRP • RFC 1075 • Similar to Routing Information Protocol (RIP) • <destinations, distances> • distance: hop count • difference • reverse-path distance • handling of tunnels • periodic exchange for the routing information • Maintain Separate Tables for Unicast & Multicast • Reverse Path Forwarding Algorithm & Pruning • first packet sent to all routers • prune message sent to back • Fixed Size IGMP Header

21. MOSPF • 패킷의 전달 경로는 패킷의 시작점과 목적지에 의해 결정됨. • OSPF를 이용 최단 경로 결정 • 그룹 구성원간에는 하나의 최단 경로만 존재 • 데이터 링크 멀티캐스팅 • Forwarding • forwading cache • Dijkstra algorithm을 이용한 최단경로 설정은 과부하 작업임 • local group database • {group A, Net1} • shortest path tree

22. MOSPF • An enhancement of the unicast routing protocol OSPF • Designed to operate within a single AS(autonomous system) • OSPF • Link-State Protocol • Complete picture of the topology of AS • MOSPF • Add Group membership-LSA • Intra-area multicasting, Inter-area multicasting, Inter-AS multicasting • Design Goals • Extend OSPF to support multicasting • Add minimal functionality to OSPF to support multicast

23. MOSPF • Protocol Data Structures • Local Group Database • Local group DB: [group, subnet] • Create group-member-LSAs (Link State Advertisement) • Forwarding Cache • [Upstream node, Downstream Interfaces (interface:hops)] • Multicasting routing capability • Inter-area Multicast forwarder

24. MOSPF • Protocol • Joining a multicast group • IGMP query • IGMP response • Create an entry in local group database • Send group membership LSA • Create an entry in local group databaseCreate a forwarding cache entry • Leaving the multicasting group

25. PIM(Protocol Independent Multicast) • IETF IDMR WG에서 현재 개발중 • Motivation: • DVMRP good for dense group membership • Need shared/source-based tree flexibility • Independence from Unicast Routing • PIM implementations do require the presence of some unicast routing protocol to provide routing table information and adapt to topology changes. • Two modes : according to the density of group members in the Internet. • Dense Mode • Sparse Mode

26. PIM-DM(2) • If a router receives a multicast packet from source S to group G, it first checks in the standard unicast routing table that the incoming interface is the one that is used for sending unicast packets toward S. If this is not the case, it drops the packets and sends back a “prune (S,G)” message on the incoming interface. • The router will then forward a copy of the message on all the interfaces for which it has not already received a “prune(S,G)”message. If there are no such interfaces, i.e., if all the interfaces have been pruned, it drops the packet and sends back a “prune(S,G)” message on the incoming interface.

27. Multiple Routers on a Broadcast Network S sends a multicast message; M is a group member. C sends a prune back, which would kill the group for M as well. Solution: The prune messages are always sent to the “all-routers” multicast address (224.0.0.2). Upon seeing the prune message, B would rejoin the group. PIM-DM(3)

28. Multipath on Broadcast Networks S sends a multicast packet to group M on E1. Both A-C and B-D routers pick it up and transmit on E2 (multiple copies). Solution: both C and D will see each other’s packet, and note that the group route points to the interface where it was received. Extension to IGMP to resolve (use shortest path) PIM-DM(4)

29. PIM-DM(5) • DVMRP와 다른점 • PIM-DM은 토폴로지의 변화를 반영하기 위해 유니캐스트 라우팅 프로토콜의 존재를 가정한다. • Child interface를 계산하지 않고 Explicit prune message가 downstream link로 부터 오기 전까지는 단순히 멀티캐스트 트래픽을 포워드 한다. • PIM-DM control message의 처리와 data packet forwarding과정은 PIM-SM과 통합되어 있어서 하나의 라우터가 다른 그룹에 대해서는 다른 mode로 작동할 수 있다.

30. PIM-SM • Dense Mode와 다른점 • 라우터가 명시적으로 Join 메시지를 보내야 한다. • Rendezvous Point를 이용한다. • Host joins a group

31. PIM-SM • Source send to a multicast group

32. PIM-SM • RP-Shared Tree or Shortest Path Tree(SPT)

33. Mobile Environment WGs in IETF Routing Area IS-IS for IP Internets (ISIS) Inter-Domain Routing (IDR) Unicast Routing Open Shortest Path First IGP (OSPF) Routing Information Protocol (RIP) UniDirectional Link Routing (UDLR) Inter-Domain Multicast Routing (IDMR) Multicast Extensions to OSPF (MOSPF) Multicast Routing Protocol Independent Multicast (PIM) Border Gateway Multicast Protocol (BGMP) Multicast Source Discovery Protocol (MSDP) IP Routing for Wireless/Mobile Hosts (MobileIP) Mobile Ad-hoc Networks (MANET) Data Link Switching MIB (DLSwMIB) General Switch Management Protocol (GSMP) Switching Multiprotocol Label Switching (MPLS) SNA DLC Services MIB (SNADLC) Virtual Router Redundancy Protocol (VRRP)

34. Routing Protocols Summary Multicast Unicast RIP DVMRP MOSPF OSPF PIM-DM ISIS PIM-SM UDLR Intra-AS CBT Long-Term Approach BGP BGMP Inter-AS SSM IDRP MSDP Short-Term Approach SGM