Skip to main content

Short note: Feedback suppression in multicast

Short answer

Feedback suppression reduces NACK implosion in reliable multicast. A receiver that misses a message waits a random time and multicasts its NACK; other receivers missing the same message suppress their own NACKs. The sender then multicasts the retransmission to the group.

Write a short note on feedback suppression mechanism in M-cast communication.

Feedback suppression is a technique in scalable reliable multicast (e.g. SRM, Scalable Reliable Multicast) for reducing the number of feedback messages (NACKs) from receivers to the sender.

If receivers each send a NACK for every lost message, a single loss at a router can cause hundreds of NACKs: a feedback implosion. In feedback suppression:

  1. A receiver that misses a message waits for a random time before sending its NACK, and multicasts the NACK to the whole group (including other receivers).
  2. Any other receiver that was going to send a NACK for the same message and now sees someone else's NACK suppresses its own.
  3. The sender gets one (or very few) NACK and multicasts the retransmission to everybody (or the nearest member who has the message can retransmit).

Advantages: fewer messages, scalable. Disadvantages: receivers which did not lose the message get the retransmission too; the random delay has to be tuned, and it needs a multicast service and may cause delay.

✓ Answer checked by Claude Opus 5.5 (high effort)

Related questions

All Distributed System chapter 9 questions →