Lightweight probabilistic broadcast
read more
Citations
Scribe: a large-scale and decentralized application-level multicast infrastructure
SplitStream: high-bandwidth multicast in cooperative environments
A survey of attack and defense techniques for reputation systems
Bullet: high bandwidth data dissemination using an overlay mesh
SCRIBE: The Design of a Large-Scale Event Notification Infrastructure
References
The many faces of publish/subscribe
Epidemic algorithms for replicated database maintenance
A reliable multicast framework for light-weight sessions and application level framing
Related Papers (5)
Frequently Asked Questions (10)
Q2. What is the effect of l on the reliability of the system?
With buffers for notifications of infinite length, as the authors have supposed in the analysis, reliability would remain constant as l becomes smaller.
Q3. What is the key concept underlying the scalability properties of gossip-based broadcast algorithms?
Decentralization is the key concept underlying the scalability properties of gossip-based broadcast algorithms, i.e., the overall load of retransmissions is reduced by decentralizing the effort.
Q4. How does the probability of a partitioning change with increasing n be reproduced?
The fact that the membership becomes more stable with an increased n can be intuitively reproduced since, with a large system, membership information becomes more sparsely distributed, and the probability of having concentrated exclusive knowledge becomes vanishingly small.
Q5. What is the nature of gossip-based broadcast protocols?
While retransmission requests in SRM can be handled by any process but lead to the re-broadcasting of a message, gossip-based protocols abide even better to the nature of peerto-peer computing, by relying on pairwise interaction between peers.
Q6. Why does the lpbcast algorithm not scale over a couple of hundred processes?
Due to the overhead of message loss detection and reparation, protocols offering such strong guarantees do not scale over a couple of hundred processes [25].
Q7. What are the main problems of network-level protocols?
Network-level protocols have turned out to be insufficient: IP multicast [6] lacks reliability guarantees, and reliable protocols do not scale well.
Q8. how to combine their membership approach with other gossip-based event dissemination algorithms?
The authors are indeed currently investigating how to combine their membership approach with other gossip-based event dissemination algorithms, e.g., using loggers to ensure strong reliability guarantees whenever this is required (cf. rpbcast).
Q9. Why is a higher fanout required to obtain similar results than with lpbcast?
In fact, because repetitions and hops are limited in the case of pbcast, a higher fanout is required to obtain similar results than with lpbcast (F = 5 here vs F = 3 in Figure 6(a)).
Q10. How many processes are likely to crash during a run?
The probability of a message loss does not exceed a predefined ε > 0, and the number of process crashes in a run does not exceed f < n.