Transcript Document
The Cost of Services in Active Reliable Multicast M. MAIMOUR, J. Mazuy and C. D. PHAM INRIA-RESO RESAM UCB-Lyon – ENS Lyon AMS’02, Edimburgh Tuesday, July 23rd, 2002 Outline • • • • • Introduction The DyRAM protocol DyRAM implementation Cost of active services in DyRAM Conclusion From unicast… Sender • Problem Sending same data to many receivers via unicast is inefficient. data data data data data data Receiver Receiver Receiver …to multicast on the Internet. Sender Problem Sending same data to many receivers via unicast is inefficient. Solution data data data data Using multicast is more efficient Receiver Receiver Receiver Reliable multicast • At the routing level : IP Multicast provides efficient delivery without any reliability guarantees. • Many multicast applications require reliability. • Reliability has to be addressed at a higher level. Reliable multicast protocols • End-to-end solutions : Only the end hosts (the source and/or the receivers) are involved. • In-network solutions : Routers are involved in the recovery process. Active routers-based solutions DyRAM : Dynamic Replier Active reliable Multicast • DyRAM is based on active services (routerassisted). • the recovery is performed from the receivers (no data cache at the routers) • A recovery tree is constructed on a per-packet basis via a replier election mechanism. • Use of NACKs combined with periodic ACKs. Active Services in DyRAM • • • • Early loss detection NACK suppression Subcast of repair packets Dynamic replier election Early loss detection service data4 A NACK is sent by the router NACKs suppression data4 only one NACK is forwarded to the source Replier election and subcast NAK 2 from link2 NAK 2 from link1 D0 DyRAM IP multicast 2 NAK 2 0 1 Repair 2 NAK 2,@ D1 Repair 2 DyRAM Repair 2 IP multicast R1 NAK 2 1 0 IP multicast NAK 2,@ NAK 2,@ IP multicast NAK 2 R4 R3 IP multicast Repair 2 R2 R5 R7 Where to place the active routers ? PSTN ISDN xDSL GSM, UMTS 10Mbits/s core network Gbits/s Server 100Mbits/s wireless LAN 1Mbits/s, 10MBits/s visio-conferencing Location of the loss detectioncapable routers The loss detection service should be located not too far from the source so the corresponding overhead is justified ! Specialized active routers architecture source The active router associated to the source can perform early processing on packets. core network Gbits rate A hierarchy of active routers can be used for processing specific functions at different layers of the hierarchy : NACK suppression, subcast, replier election. Simulation model Simulation results 4 receivers/group p=0.25 #grp: 6…24 DyRAM implementation Preliminary experimental results Testbed configuration • TAMANOIR active execution environment • Java 1.3.1 and a linux kernel 2.4 • A set of PCs receivers and 2 PC-based routers ( Pentium II 400 MHz 512 KB cache 128MB RAM) • Data packets are of 4KB Packets format : ANEP • Data /Repair packets S@IP D@IP SVC • NACK SEQ isR packets S@IP D@IP SVC SEQ S@IP Payload The Tamanoir execution environment FTP TAMANOIR S S1 S,@IP FTP port data Tamanoir port UDP IP IP UDP S,@IP data ANEP packet Router’s data structures • The track list TL which maintains for each multicast session, • lastOrdered : the sequence number of the last received packet in order • lastReceived : the sequence number of the last received data packet • lostList : list of not received data packets in between. • The Nack structure NS that keeps for each lost data packet, • seq : sequene number of the lost data packet • subList : list of IP addresses of the downstream receivers (active routers) that have lost it. The first configuration ike resama resamo stan resamd Active services cost •NACK : 135μs •DP : 20μs if no seq gap, 12ms-17ms otherwise. Only 256μs without timer setting •Repair : 123μs The second configuration ike resamo The replier election cost •The election is performed on-the-fly. •It depends on the number of downstream links. •ranges from 0.1 to 1ms for 5 to 25 links per router. Conclusion and future work • The results of the processing cost are promising (tens to hundreds microsec.) • But the overhead due to setting timers should be addressed (many optimizations are possible) • Optimizing the replier election based on an estimation of the receivers power (by means of BW, delay …) • A congestion control is currently under evaluation and will be integreted into DyRAM in the near future.