## Randomized rumor spreading (2000)

### Cached

### Download Links

- [cone.informatik.uni-freiburg.de]
- [archive.cone.informatik.uni-freiburg.de]
- [www.mpi-sb.mpg.de]
- [www.itheoi.mu-luebeck.de]
- DBLP

### Other Repositories/Bibliography

Venue: | In IEEE Symposium on Foundations of Computer Science |

Citations: | 197 - 1 self |

### BibTeX

@INPROCEEDINGS{Karp00randomizedrumor,

author = {R. Karp and C. Schindelhauer and S. Shenker and B. Vöcking},

title = {Randomized rumor spreading},

booktitle = {In IEEE Symposium on Foundations of Computer Science},

year = {2000},

pages = {565--574}

}

### Years of Citing Articles

### OpenURL

### Abstract

We investigate the class of so-called epidemic algorithms that are commonly used for the lazy transmission of updates to distributed copies of a database. These algorithms use a simple randomized communication mechanism to ensure robustness. Suppose players communicate in parallel rounds in each of which every player calls a randomly selected communication partner. In every round, players can generate rumors (updates) that are to be distributed among all players. Whenever communication is established between two players, each one must decide which of the rumors to transmit. The major problem (arising due to the randomization) is that players might not know which rumors their partners have already received. For example, a standard algorithm forwarding each rumor from the calling to the called players for rounds needs to transmit the rumor times in order to ensure that every player finally receives the rumor with high probability. We investigate whether such a large communication overhead is inherent to epidemic algorithms. On the positive side, we show that the communication overhead can be reduced significantly. We give an algorithm using only transmissions and rounds. In addition, we prove the robustness of this algorithm, e.g., against adversarial failures. On the negative side, we show that any address-oblivious algorithm (i.e., an algorithm that does not use the addresses of communication partners) needs to send messages for each rumor regardless of the number of rounds. Furthermore, we give a general lower bound showing that time- and communicationoptimality cannot be achieved simultaneously using random phone calls, that is, every algorithm that distributes a rumor

### Citations

718 |
Epidemic algorithms for replicated database maintenance
- Demers, Greene, et al.
- 1987
(Show Context)
Citation Context ...tion whether this additional communication effort is a special property of the above push algorithm or is inherent to rumor spreading using random phone calls in general. 1.1 Background Demers et al. =-=[2]-=- introduced the idea of using so-called epidemic algorithms for the lazy update of data objects in a data base replicated at many sites, e.g., yellow pages, name servers, or server directories. In par... |

272 |
A survey of gossiping and broadcasting in communication networks
- Hedetniemi, Hedetniemi, et al.
- 1988
(Show Context)
Citation Context ...an also inform all players in O(ln n) rounds using a deterministic interconnection of constant degree, e.g., a shuffle network. (For an overview of deterministic information dissemination we refer to =-=[5]-=- or [6].) The advantage of the randomized push algorithm, however, is its inherent robustness against several kinds of failures compared to deterministic schemes that either need substantially more ti... |

140 | Providing high availability using lazy replication
- Ladin, Liskov, et al.
- 1992
(Show Context)
Citation Context ...iated an enormous amount of experimental and conceptual study of epidemic algorithms. For example, there is a variety of research issues like consistency, correctness, data structures, and efficiency =-=[1, 7, 8, 9, 12]-=-. Recent theoretical work concentrates on the robustness against Byzantine failures [11]. In this paper, we concentrate only on the efficiency of these randomized algorithms. In particular, we study t... |

99 | Dissemination of information in interconnection networks (broadcasting and gossiping
- Hromkovic, Klasing, et al.
- 1993
(Show Context)
Citation Context ... inform all players in O(ln n) rounds using a deterministic interconnection of constant degree, e.g., a shuffle network. (For an overview of deterministic information dissemination we refer to [5] or =-=[6]-=-.) The advantage of the randomized push algorithm, however, is its inherent robustness against several kinds of failures compared to deterministic schemes that either need substantially more time [4] ... |

72 |
Epidemic algorithms in replicated databases
- Agrawal, Abbadi, et al.
- 1997
(Show Context)
Citation Context ...iated an enormous amount of experimental and conceptual study of epidemic algorithms. For example, there is a variety of research issues like consistency, correctness, data structures, and efficiency =-=[1, 7, 8, 9, 12]-=-. Recent theoretical work concentrates on the robustness against Byzantine failures [11]. In this paper, we concentrate only on the efficiency of these randomized algorithms. In particular, we study t... |

45 | On the fault tolerance of some popular bounded-degree networks
- Leighton
- 1992
(Show Context)
Citation Context ...r, is its inherent robustness against several kinds of failures compared to deterministic schemes that either need substantially more time [4] or can tolerate only a relatively small number of faults =-=[10]-=-. For example, consider node failures in which a player (different from the player starting the rumor) fails to communicate or simply crashes and forgets its rumors. Obviously, when using a sparse det... |

36 | Scalable update propagation in epidemic replicated databases
- Rabinovich, Gehani, et al.
- 1996
(Show Context)
Citation Context ...iated an enormous amount of experimental and conceptual study of epidemic algorithms. For example, there is a variety of research issues like consistency, correctness, data structures, and efficiency =-=[1, 7, 8, 9, 12]-=-. Recent theoretical work concentrates on the robustness against Byzantine failures [11]. In this paper, we concentrate only on the efficiency of these randomized algorithms. In particular, we study t... |

28 | T.W. Consistency algorithms for optimistic replication
- Guy, Popek, et al.
- 1993
(Show Context)
Citation Context |

28 | On Diffusing Updates in a Byzantine Environment
- Malkhi, Reiter
- 1999
(Show Context)
Citation Context ...here is a variety of research issues like consistency, correctness, data structures, and efficiency [1, 7, 8, 9, 12]. Recent theoretical work concentrates on the robustness against Byzantine failures =-=[11]-=-. In this paper, we concentrate only on the efficiency of these randomized algorithms. In particular, we study their time and communication complexity using a simple model for the underlying randomize... |

18 | Accessing replicated data in a large-scale distributed system
- Golding, Long
- 1991
(Show Context)
Citation Context |

16 | Adaptive broadcasting with faulty nodes
- Gasieniec, Pelc
- 1996
(Show Context)
Citation Context ... [6].) The advantage of the randomized push algorithm, however, is its inherent robustness against several kinds of failures compared to deterministic schemes that either need substantially more time =-=[4]-=- or can tolerate only a relatively small number of faults [10]. For example, consider node failures in which a player (different from the player starting the rumor) fails to communicate or simply cras... |

5 |
and bins: A study in negative dependence
- Balls
- 1996
(Show Context)
Citation Context ...ependent since the random interconnections used for transmissions in phase i form partial permutations on the caller sites. This dependence, however, is negative so that we can apply a Chernoff bound [3]. Assuming u(i) (ln n) 2 , we obtain Pr [U isu(i)] Pr U is(1 1 3 )E [U(i)] exp( 1 18 E [U(i)]) exp( 1 12 u(i)) = O(n ) ; for any positive constant . This completes the proof of Theorem ... |