Baruch Awerbuch, Israel Cidon, and Shay Kutten. Optimal maintenance of replicated information. In Proc. 31st IEEE Symp. on Foundations of Computer Science, 1990.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....important example is the case that the replicated information is the set of non faulty links adjacent to each node. Maintaining replicas of this information is the classical Topology Update problem, where each node is required to know the description of its connected component of the network [16, 6, 11, 14, 7, 3]. This is one of the most common tasks performed in existing networks since, when the topology gets to be known to all nodes, many distributed tasks can be reduced to sequential tasks. This reduction is conducted by having each node simulate the distributed task on the topology known to it. An ....

....be known to all nodes, many distributed tasks can be reduced to sequential tasks. This reduction is conducted by having each node simulate the distributed task on the topology known to it. An Example for a protocol which uses this approach is the Internet OSPF interior routing protocol [12] In [3] it was shown that the incremental cost of adapting to a single topology change can be smaller than the communication complexity of the previous approach [1] of solving the problem from scratch . A variant of the algorithm of [3] was implemented later as a part of the PARIS networking project at ....

[Article contains additional citation context not shown here]

Baruch Awerbuch, Israel Cidon, and Shay Kutten. Optimal maintenance of replicated Information. Proceedings of the 31st Annual IEEE Symposium on Foundations of Computer Science (FOCS 90), St. Louis, MO, USA, pp.492-- 502, October 1990.

....at a cost much smaller than restarting the task. In particular, our protocol recovers from failures with a small fixed cost, independent of the number of broadcasts done so far. There have been several previous works which featured a smaller cost per recovery than the blast away approach [ACK90] Awe88] AS88] BGS88] MS79] SS81] We therefore propose another simplified form for the communication complexity. This communication complexity consists of fixed cost per each accept, failure and recovery event. The cost depends on the kind of the event. Definition 10 Consider functions ....

Baruch Awerbuch, Israel Cidon, and Shay Kutten. Optimal maintenance of replicated information. In Proc. 31 st Symp. on Foundations of Computer Science, October 1990.

....the crash of a single node, hence their worst case complexity is not very attractive. However, when the number of changes is small, the update approach may lead to considerable savings in communication for many problems, since most of the information does not need to be redistributed. It is argued [ACK90] that the reset approach can be avoided, since other approaches can be made more efficient even in the worst case. In [ACK90] a spanning tree is maintained rather than recomputed. When a tree edge fails, 5 the algorithm replaces it by a single other tree edge (if possible) keeping the rest of ....

....the update approach may lead to considerable savings in communication for many problems, since most of the information does not need to be redistributed. It is argued [ACK90] that the reset approach can be avoided, since other approaches can be made more efficient even in the worst case. In [ACK90], a spanning tree is maintained rather than recomputed. When a tree edge fails, 5 the algorithm replaces it by a single other tree edge (if possible) keeping the rest of the tree intact. This property is called there path preservation, and it is argued that it keeps the routing on the tree ....

Baruch Awerbuch, Israel Cidon, and Shay Kutten. Optimal maintenance of replicated information. In Proc. 31st IEEE Symp. on Foundations of Computer Science, 1990.

....in new high speed networks such as IBM s PARIS [CG88, ACG 90] and NSF DARPA s AURORA gigabit rate testbed. The update protocols are used to handle management tasks such as routing and bandwidth reservations. As further motivation for the study of the Topology Update problem, it was shown in [ACK90] that it is in fact the hardest instance of a database update problem. Note also that topology dependent tasks such as Shortest Paths and Minimum Spanning Tree can be solved locally in a network in which current replicas of the topology are maintained at each node. Of course individual problems ....

....as Full Broadcast, can be very wasteful in communication since it fails to take advantage of prior partial knowledge available in the system. For, processors need be informed of relatively few changes if they already hold a nearly correct view of the database. The Incremental Update strategy in [ACK90, ACG 90] exploits fully the prior knowledge available in the system, sending each processor only one message per error. While Incremental Update is clearly superior to Full Broadcast in terms of communication complexity, it can be significantly inferior in terms of time complexity. This is ....

[Article contains additional citation context not shown here]

Baruch Awerbuch, Israel Cidon, and Shay Kutten. Optimal maintenance of replicated information. In Proc. 31st IEEE Symp. on Foundations of Computer Science, 1990.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC