Multicast Review 2 v Differences between Ethernet and

Multicast

Review 2 v Differences between Ethernet and IP v Differences between forwarding and routing v Differences between link-state routing and Dijstra’s algorithm Hui Zhang

Multicast Routing 3 v Unicast: one source to one destination v Multicast: one source to many destinations v Main goal: efficient data distribution Hui Zhang

Multicast – Efficient Data Distribution Src 4 Src Hui Zhang

Example Applications 5 v Broadcast audio/video v Push-based systems v Software distribution v Web-cache updates v Teleconferencing (audio, video, shared whiteboard, text editor) v Multi-player games v Server/service location v Other distributed applications Hui Zhang

IP Multicast Architecture Service model Hosts Host-to-router protocol (IGMP) Routers Multicast routing protocols (various) 6 Hui Zhang

Logical Naming v Single name/address maps to logically related set of destinations § v Key challenge: scalability § 7 Destination set = multicast group Single name/address independent of group growth or changes Hui Zhang

Multicast Router Responsibilities v Learn of the existence of multicast groups (through advertisement) v Identify links with group members v Establish state to route packets § Replicate packets on appropriate interfaces § Routing entry: Src, incoming interface, group address 8 List of outgoing interfaces Hui Zhang

IP Multicast Service Model (rfc 1112) v Each group identified by a single IP address v Groups may be of any size v Members of groups may be located anywhere in the Internet v Members of groups can join and leave at will v Senders need not be members v Group membership not known explicitly v Analogy: § 9 Each multicast address is like a radio frequency, on which anyone can transmit, and to which anyone can tune-in. Hui Zhang

IP Multicast Addresses v Class D IP addresses § 224. 0. 0. 0 – 239. 255 1 110 v 10 Group ID How to allocated these addresses? § Well-known multicast addresses, assigned by IANA § Transient multicast addresses, assigned and reclaimed dynamically, e. g. , by “sdr” program Hui Zhang

IP Multicast API 11 v Sending – same as before v Receiving – two new operations § Join-IP-Multicast-Group(group-address, interface) § Leave-IP-Multicast-Group(group-address, interface) § Receive multicast packets for joined groups via normal IPReceive operation § Implemented using socket options Hui Zhang

Multicast Scope Control – Small TTLs v TTL expanding-ring search to reach or find a nearby subset of a group s 1 2 3 12 Hui Zhang

Multicast Scope Control – Large TTLs v Administrative TTL Boundaries to keep multicast traffic within an administrative domain, e. g. , for privacy or resource reasons The rest of the Internet An administrative domain 13 TTL threshold set on interfaces to these links, greater than the diameter of the admin. domain Hui Zhang

IP Multicast Architecture Service model Hosts Host-to-router protocol (IGMP) Routers Multicast routing protocols (various) 14 Hui Zhang

Internet Group Management Protocol v End system to router protocol is IGMP v Each host keeps track of which mcast groups are subscribed to § v Objective is to keep router up-to-date with group membership of entire LAN § 15 Socket API informs IGMP process of all joins Routers need not know who all the members are, only that members exist Hui Zhang

How IGMP Works Routers: Q Hosts: 16 v On each link, one router is elected the “querier” v Querier periodically sends a Membership Query message to the allsystems group (224. 0. 0. 1), with TTL = 1 v On receipt, hosts start random timers (between 0 and 10 seconds) for each multicast group to which they belong Hui Zhang

How IGMP Works (cont. ) Routers: Hosts: 17 Q G G v When a host’s timer for group G expires, it sends a Membership Report to group G, with TTL = 1 v Other members of G hear the report and stop their timers v Routers hear all reports, and time out non-responding groups Hui Zhang

How IGMP Works (cont. ) v Note that, in normal case, only one report message per group present is sent in response to a query § 18 Power of randomization + suppression v Query interval is typically 60 -90 seconds v When a host first joins a group, it sends one or two immediate reports, instead of waiting for a query Hui Zhang

IP Multicast Architecture Service model Hosts Host-to-router protocol (IGMP) Routers Multicast routing protocols (various) 19 Hui Zhang

Routing Techniques v Basic objective – build distribution tree for multicast packets v Flood and prune v 20 § Begin by flooding traffic to entire network § Prune branches with no receivers § Examples: DVMRP, PIM-DM § Unwanted state where there are no receivers Link-state multicast protocols § Routers advertise groups for which they have receivers to entire network § Compute trees on demand § Example: MOSPF § Unwanted state where there are no senders Hui Zhang

Routing Techniques v 21 Core based protocols § Specify “meeting place” aka core § Sources send initial packets to core § Receivers join group at core § Requires mapping between multicast group address and “meeting place” § Examples: CBT, PIM-SM Hui Zhang

Shared vs. Source-based Trees v v 22 Source-based trees § Separate shortest path tree for each sender • DVMRP, MOSPF, PIM-DM, PIM-SM Shared trees § Single tree shared by all members § Data flows on same tree regardless of sender § CBT, PIM-SM Hui Zhang

Source-based Trees Router S Source R Receiver R R S R 23 Hui Zhang

Shared Tree Router S Source R Receiver R R S RP R S R 24 Hui Zhang

Shared vs. Source-Based Trees v v v 25 Source-based trees § Shortest path trees – low delay, better load distribution § More state at routers (per-source state) § Efficient for in dense-area multicast Shared trees § Higher delay (bounded by factor of 2), traffic concentration § Choice of core affects efficiency § Per-group state at routers § Efficient for sparse-area multicast Which is better? Hui Zhang

Multicast OSPF (MOSPF) 26 v Add-on to OSPF (Open Shortest-Path First, a link-state, intra-domain routing protocol) v Multicast-capable routers flag link state routing advertisements v Link-state packets include multicast group addresses to which local members have joined v Routing algorithm augmented to compute shortest-path distribution tree from a source to any set of destinations Hui Zhang

Example Source 1 Z W Q T Receiver 1 Receiver 2 27 Hui Zhang

Link Failure/Topology Change Source 1 Z W Q T Receiver 1 Receiver 2 28 Hui Zhang

Membership Change Source 1 Z Receiver 3 W Q T Receiver 1 Receiver 2 29 Hui Zhang

Impact on Route Computation 30 v Can’t pre-compute multicast trees for all possible sources v Compute on demand when first packet from a source S to a group G arrives v New link-state advertisement § May lead to addition or deletion of outgoing interfaces if it contains different group addresses § May lead to re-computation of entire tree if links are changed Hui Zhang

Distance-Vector Multicast Routing v v 31 DVMRP consists of two major components: § A conventional distance-vector routing protocol (like RIP) § A protocol for determining how to forward multicast packets, based on the routing table DVMRP router forwards a packet if § The packet arrived from the link used to reach the source of the packet (reverse path forwarding check – RPF) § If downstream links have not pruned the tree Hui Zhang

Example Topology G G S G 32 Hui Zhang

Broadcast with Truncation G G S G 33 Hui Zhang

Prune G G Prune (s, g) S Prune (s, g) G 34 Hui Zhang

Graft G G G Report (g) Graft (s, g) S Graft (s, g) G 35 Hui Zhang

Steady State G G G S G 36 Hui Zhang
- Slides: 36