Multicast and Broadcast for the Future Internet
1/ 34
the Future Internet
Hojin Lee ([email protected])
2008.02.20
Contents• Multicast Research
– Native IP Multicast– Application Layer Multicast
• Main Driving Forces
2/ 34
• Main Driving Forces• Indirection• Conclusion
Multicast Research
3/ 34
Native IP Multicast• Based on IP Network• Many intra-domain, inter-domain multicast routing protocols– Intra-domain: DVMRP, MOSPF, CBR, PIM-DM, PIM-SM, …
4/ 34
SM, …– Inter-domain: QoSMIC, PIM-SM/MSDP, BGMP, EXPRESS, …
• Support large multicast group• However, large number of multicast groups?
Application Layer Multicast• To address deployability problem of native IP multicast– POM (Proxied Overlay Multicast)– ESM (End System Multicast)
5/ 34
– ESM (End System Multicast)
• Efficiency?
Classification
6/ 34
Illustration
7/ 34
(a): Native IP multicast (b): ESM (c): POM
Activities of Multicast Research (1/2)
• IETF WG– multicast group management
• magma WG (http://www.ietf.org/html.charters/magma-charter.html)
– multicast routing protocols
8/ 34
– multicast routing protocols• mboned WG (http://www.ietf.org/html.charters/mboned-
charter.html)• pim WG (http://www.ietf.org/html.charters/pim-charter.html)• idr WG (http://www.ietf.org/html.charters/idr-charter.html)
– missing features in current multicast• rmt WG (http://www.ietf.org/html.charters/rmt-charter.html)• msec WG (http://www.ietf.org/html.charters/msec-charter.html)
Activities of Multicast Research (2/2)
• IETF RG– Scalable Adaptive Multicast RG (http://www.samrg.org)
9/ 34
to introduce broadcast and multicast
Main Driving Forces10
10/ 34
to introduce broadcast and multicast schemes…
L. Henden, et al., “Broadcast and multicast – a vision on their role in future broadband access networks,” Broacast Multicast Cluster, IST, Jan 2005
1. Saving Bandwidth• Effective utilization
of bandwidth– Bandwidth and server requirement will increase fast.
11
11/ 34
increase fast.
– Current solution: increase the capacity of both the backbone and the server at the same speed. � A cost-intensive concealment of the real problem
2. Economy12
• The costs of multicast services are related to:– Network deployment– Installation of CPE– Management– Maintenance
• For multicast,
12/ 34
• For multicast,– The initial costs are higher– The cost of adding new users decreases proportional with
increasing number of receivers.• The use of multicast only make sense when the bandwidth savings are larger than the deployment and management cost.
3. Introduction of E-based Services
• Participation in the modern e-society– Digital divide
• It represents a threat to the development of an information society serving everybody.
13
13/ 34
• Community type services– E-government, E-health, E-learning– A huge amount of information should be
delivered to many receivers, which will take advantage of multicast
(4. Changes of Traffic Type)• Data -> Streaming
• Live streaming service– Real time sport broadcasting
14/ 34
– Real time sport broadcasting– News– …
• IPTV, …
Indirection
15/ 34
Unicast vs. Multicast• IP unicast
– direction: a destination address• IP multicast
– indirection: a group address instead of individual
R1
R2
RG
16/ 34
– indirection: a group address instead of individual destination addresses
* Indirection: the ability to reference something using a name, reference,
or container instead of the value itself [ wikipedia]
R3
R4
R5
G
Basic• Indirection architecture
– Each packet is associated with an id – Receiver: inserts trigger (id, addr) into the network– Sender: sends packet (id, data)
17/ 34
– support general communication abstractions• unicast, multicast, anycast, host mobility
General Communication Abstractions
• multicast
• mobility
18/ 34
• anycast
Multicast• Receivers insert triggers with same identifier• Can dynamically switch between multicast and unicast
19/ 34
Sender Receiver (R1)
Receiver (R2)
trigger
id R2
trigger
id R1
Mobility• Host just needs to update its trigger as it moves from one subnet to another
20/ 34
Sender
Receiver
(R1)
Receiver
(R2)
id R1id R2
Anycast• Use longest prefix matching instead of exact matching– Prefix p: anycast group identifier– Suffix si: encode application semantics, e.g., location
send(R1,data)
21/ 34
Sender
Receiver (R1)
p|s1 R1send(p|a,data)
Receiver (R2)p|s2 R2
p|s3 R3
Receiver (R3)
Underlay vs. Overlay• Underlay routing (network infra entity)
– simple for scalability, efficiency, cost and robustness• unicast
22/ 34
• unicast– general communication abstractions are supported by overlay• can be applied into a new network with minor or no modification
Example• ID space [0..63] partitioned across five i3 nodes • Each host knows one i3 node• R inserts trigger (37, R); S sends packet (37, data)
3S
send(37, data)
02m-1
23/ 34
7
20
35
41
37 R
3
7
20
35
41
37 R
R
trigger(37,R)
send(R, data)
Chord circle
R
Multicast• Logical links between indirection servers– bound out-degree
24/ 34
• Tree optimization
Indirection Multicast Architecture• Hierarchy
– physical topology consideration• Scalability
– member aggregation
25/ 34
– member aggregation– flow aggregation
• User indirection multicast
Hierarchy• Assign ID to servers
– randomly• flat label
– with additional information
26/ 34
– with additional information• hierarchy
– more hierarchy, less anonymity• why?
– aggregation
Aa0 Aa1
Ab0
Ac0Ac1
Ba0
Ba1
source
aggregation
Infra Indirection Multicast
Big Picture
27/ 34
Aa1
Aa2
Ac2
Ba2Bb1
Bb0
User Indirection Multicast
Ba0 Ba1
HierarchyBa0Ab0
source
28/ 34
Aa0 Aa1
Aa2
HierarchyBa0 Ba1
Ab0
source
29/ 34
Aa0 Aa1
Aa2
Hierarchy - Issues• How to make hierarchical ID for indirection servers– ?: ASN# + IP prefix + random value
• Which ID is assigned to a new
30/ 34
• Which ID is assigned to a new indirection server– ?: two neighbor indirection servers’ ID / 2
• Reassignment indirection server ID– ?
Scalability• IP unicast
– LPM -> aggregation• IP multicast
– multicast group addresss has no meaning (flat label)
31/ 34
– multicast group addresss has no meaning (flat label) -> no-aggregation => state per flow
– cf> volatile– scalable in terms of the number of multicast members
– not scalable in terms of the number of multicast groups
Scalability – flow aggregation• ID assignment
– Flexible ID assignment and re-assignment for a flow
– An ID assignment for a group of flowsabc.001
abc.xxx
32/ 34
– An ID assignment for a group of flows
• ID summarization– ID & group_prefix = a set of flows
abc.001
abc.002
abc.003
abc.999
User Indirection Multicast• Dense multicast group in the same subnet or nearby subnet– Temporarily end host performs indirection– end host
33/ 34
– end host• less powerful, limited uplink bandwidth• less stable• my resource
Conclusion• Using Indirection
– general communication abstractions– anonymity
• Focus on multicast
34/ 34
• Focus on multicast– physical topology consideration– scalability– user indirection
References• i3• wikipedia• secure indirection• chord
35/ 34
• chord
Appendix• Splitting L3 into
– communication service layer
– forwarding layer communication service layer
- multicast
APP/ transport
36/ 34
– forwarding layer - multicast
- anycast
- mobility
MAC/ PHY
forwarding layer
(basic: unicast)
Aa0Aa1
Aa2
Bb0
Bb1
Bb2
37/ 34
Ac2
Ba0
Ba1
Ab0
Ac1
Ac0