1 / 15

Reliable Multicast and Layered Multicast

This review discusses the application layer multicast protocol, Narada, and the challenges of reliable multicast. It explores the Scalable Reliable Multicast (SRM) algorithm and proposes adaptive timers. The review also examines the issues of sending data to receivers and presents two multi-rate schemes: Destination Set Grouping (DSG) and Layered Approach.

scamp
Download Presentation

Reliable Multicast and Layered Multicast

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Reliable Multicast and Layered Multicast 12/3/2001

  2. Review: Application Layer Multicast • Narada • first builds a mesh • then runs DVMRP over the mesh • Overcast • directly builds a tree (the non-leaf nodes are servers) • when connecting to the tree, a new node probes from leaf to root

  3. Reliable Multicast: Discussion • What are the goals of reliable multicast? • Why is reliable multicast difficult?

  4. Scalable Reliable Multicast (SRM) [Floyd et al. ’95] • Randomize NACKs (requests) • All traffic including requests and repairs are multicast • A repair can be sent by any node that heard the request • A node suppresses its request if another node has just sent a request for the same data item • A node suppresses a repair if another node has just sent a repair

  5. Estimating Distances • Every node estimates distance (in time) from every other node • why? • how?

  6. Request Timer • Chosen from the uniform distribution on • A – node that lost the packet • S – source • C1, C2 – algorithm parameters • If A receives a request before its timeout triggers, it does an exponentially backoff

  7. Repair Timer • Chosen from the uniform distribution on • A – node that lost the packet • B – node that sends the repair • D1, D2 – algorithm parameters • If B receives a repair for missing data it cancels its timer • B does not verify whether A has received data (why?)

  8. Chain Topology • C1 = D1= 1, C2 = D2= 0 (why?) • All link distances are 1 source L2 L1 R1 R2 R3 data out of order data/repair request request repair request TO repair repair TO

  9. Star Topology • Choose C1 = D1= 0 (why?) • Given C2, what is the expected number of requests, expected time to receive a repair? source N1 N2 Ng N3 N4

  10. Bounded Degree Tree • Use both • deterministic suppression (chain topology) • probabilistic suppression (star topology) • Large C2/C1 fewer duplicate requests, but larger repair time • Large C1 contribute to suppressing more duplicate requests • Small C1 smaller repair time • Simulation parameters: C1 = C2= 2, D1 = D2= log10g

  11. Discussion: Adaptive Timers • How would you adapt the timers?

  12. Discussion • How scalable is SRM? • Some other ways you can think of to provide reliable multicast?

  13. Remaining Issue: How Fast to Send to the Receivers? • Why is the problem difficult?

  14. Two Multi-rate Schemes Destination Set Grouping (DSG) [JAZ95] • sender sends to K groups • sender sends to group i at gi • each receiver joins one group Layered approach [Sha92, McC96] • sender sends to K groups • assume g1 < g2 < … < gK • sender sends at g1, g2 – g1, …, gK – gK-1 • receivers that can receive gi join layers 1, …, i

  15. Layered Multicast • A receiver: • if no loss when joins layers 1 to i, try i+1; • if high loss rate when joins layers 1 to i, drops layer i to see the effects

More Related