Supporting IP Multicast Integrated Services in ATM Networks L. - - PDF document

1

Supporting IP Multicast Integrated Services in ATM Networks

• L. Salgarelli and A. Corghi, CEFRIEL

{salga, corghi}@cefriel.it

• D. Witaszek and H. Sanneck, GMD Fokus

{witaszek, sanneck}@fokus.gmd.de

2

Overview

• Introduction
• IP Multicast over ATM: EARTH
• Internet Integrated Services over ATM: Issues
• Integrated Approach: Multicast Integration Server (MIS)
• Service model: quantized heterogeneity
• RSVP server
• RSVP state modification at the MIS
• Sequence of events to establish a QoS point-to-multipoint VC
• Conclusions

3

IP Multicast over ATM: EARTH

• IP Multicast / ATM address resolution for Multicast LIS: shortcuts
• support of source-specific groups (IGMPv3) and Layer 2 QoS information

LIS B LIS A Multipoint data VC Point-to-point control VC ATM cloud (MLIS) EARTH Server EARTH_join EARTH_multi Sender Receiver 1 Receiver 2 Receiver 3 Receiver 4 EARTH_request

4

Internet Integrated Services (IIS) over ATM: Issues

• service class mapping

solved issue (e.g. IIS Guaranteed service to ATM CBR/rt-VBR)

• Receiver heterogeneity:

limited (VC space consumption, data duplication)

• dynamic QoS renegotiation:

end-to-end VC setup necessary

• RSVP merging:
• nly at edges of the ATM cloud
• shortcuts:

transport of ATM addresses with RSVP, RSVP messages still travel hop-by-hop

5

Point-to-point control VC EARTH & RSVP protocol messages LIS B LIS A Multicast Integration Server Sender Receiver 1 Receiver 2 Receiver 3 Receiver 4 ATM cloud (MLIS) Multipoint data VC

Integrated approach: Multicast Integration Server (MIS)

• Joint client/server operation of RSVP and EARTH protocols within MLIS

6

LIS B LIS A Sender Receiver 1 Receiver 4 Receiver 5 Receiver 6 ATM cloud (MLIS) Receiver 2 Receiver 3 Best-effort QoS level 1 QoS level 2

Service model: Quantized Heterogeneity

• support for a limited number of QoS levels

(QoS level: pre-defined set of QoS parameters)

7

Best-effort VC Base-Layer VC (QoS-0) Enhan.-Layer-1 VC (QoS-1) Rec.7 (QoS-0) Sender Rec.6 (QoS-0) Rec.5 (QoS-0) Rec.4 (QoS-0 + QoS-1) Rec.3 (QoS-0 + QoS-1) Rec.2 (best-effort) Rec.1 (best-effort) MLIS ATM pt-mpt VCs

Quantized heterogeneity: support for layered flows

8 LIS B LIS A Multicast Integration Server PATH messages RESV messages ATM cloud (MLIS) Sender Receiver 1 Receiver 2 Receiver 3 Receiver 4 Multipoint data VC Point-to-point control VC

RSVP operation within the MLIS: client/server

• efficient PATH state distribution, RESV merging
• interface to EARTH QoS capabilities

9

Multicast Integration Server PATH message with a single Traffic Description parameter PATH message with the original Traffic Descriptor plus the allowed QoS levels Sender Receiver 1 Receiver 2 Receiver 3 Receiver 4 PATH messages Multipoint data VC Point-to-point control VC

RSVP operation: state modification at the MIS

Example: introducing the quantized heterogeneity model

• Any administrative policy can be enforced at the MIS

10

Sequence of events to establish a pt-mpt VC

Layer-3 signalling Layer-2 signalling Layer-2 data

(Re-)Sender

1: PATH 6B: EARTH_qos_notify 5: EARTH_m

ulti

2: PATH (ADSPEC) 3: RESV (receiver)

MIS

RSVP server EARTH server

4: QSSI

Receiver

Control VC

0: EARTH_join 7: RESV (M

IS) Data (Q

• S) VC

6A

: Capacity admission control / Q

• S VC setup

8: RESV

RSVP vif EARTH control endpoint

8: RESV_ERRO

R (MIS)

4/6: RESV_ERRO

R (receiver)

• RSVP/EARTH interface only at ther server: QSSI (4)
• remote Layer 2 capacity admission control (5, 6) = ATM VC setup

11

Conclusions

• Shortcut support for multicast flows (minimizes Layer 3 processing and

resending identical data through the switch due to LIS boundaries)

• joint operation of Layer 2 (EARTH) and Layer 3 protocol (RSVP),

however with a strict functional separation

• RSVP: no changes to protocol semantics

(additional protocol objects, modified Traffic Control Interface, centralized merging point: reduced protocol overhead, policy enforcement)

• Quantized heterogeneity model:

supports scalability (VC space, data duplication) and layered flows

• Very large clouds: need to setup multiple servers with SCSP coordination
• Future work: interworking to Multilayer Routing clouds