L. Rodrigues and P. Verissimo, "Causal separators and topological timestamping an approach to support causal multicast in large-scale systems", In the Proceedings of the 15 International Conference on Distributed Computing Systems, Vancouver, British Columbia, Canada, May 1995.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....distribution of autonomous enterprise systems and the large number of hosts, could give rise to scalability issues. At the best of our knowledge, group toolkits have been used only in department (i.e. small scale) local systems. Only a few prototypes ( 5] 17] and some theoretical papers ( 6] [24]) address the problem of using a group toolkit over a large scale distributed system. On the other hand, IIOP based systems could face heterogeneity exploiting the high wrapping capabilities provided by CORBA. The main drawback resulting from their above the ORB design is performance: using ....

L. Rodrigues, P. Verissimo: "Causal Separators and Topological Timestamping: an Approach to Support Causal Multicast in Large-Scale Systems". Proceedings of the 15th International Conference on Distributed Systems, 1995.

....Moreover, both the geographical distribution of CISs and the large number of hosts, can give rise to scalability issues. To the best of our knowledge, group toolkits have been used only in departmental (i.e. small scale) local systems. Only a few prototypes [34, 35] and some theoretical papers [36, 37] address the problem of using a group toolkit over a large scale distributed system. On the other hand, IIOP based systems can deal with heterogeneity by exploiting the high wrapping capabilities provided by CORBA. The main drawback resulting from their above the ORB design is performance: ....

....of dependable integration platform. ACKNOWLEDGEMENTS The authors would like to thank Massimo Mecella, for the relevant suggestions and interesting discussions, and Raffaella Panella, Paolo Papa, Alessandro Termini, Sara In its most recent version the DOORS design team has adopted the TAO ORB [37] as the underlying ORB platform. In our opinion, though this design choice provides a performance improvement, at the same time it limits the interoperability of the DOORS platform. Tucci Piergiovanni and Luigi Verde, for their valuable contributions to the IRL project. This work was partially ....

L. Rodrigues and P. Verissimo, "Causal separators and topological timestamping: an approach to support causal multicast in large-scale systems," in Proceedings of the 15 International Conference on Distributed Systems (ICDCS '95), 1995, pp. 83-91.

....both the geographical distribution of CISs and the large number of hosts, could give rise to scalability issues. At the best of our knowledge, group toolkits have been used only in department (i.e. small scale) local systems. Only a few prototypes ( 34] 35] and some theoretical papers ( 36] [37]) address the problem of using a group toolkit over a large scale distributed system. On the other hand, IIOP based systems could face heterogeneity exploiting the high wrapping capabilities provided by CORBA. The main drawback resulting from their above the ORB design is performance: using ....

L. Rodrigues, P. Verissimo, "Causal Separators and Topological Timestamping: an Approach to Support Causal Multicast in Large-Scale Systems" in Proceedings of the 15 International Conference on Distributed Systems, 1995.

....in the same domain and summarizes information about other domains. But this technique is specially adapted to update propagation in replicated databases using a weak consistency algorithm 2 and is not suitable for causal communication. An original solution for causal delivery is introduced in [11]. This solution does not use a logical clock and implements the causal history relation with lists of causally linked messages. The nodes interconnection graph is split in subnets separated by vertex separators. This approach allows cycles in the subnet interconnection graph and reduces the size ....

....[1] Causality is only maintained on each individual bus, also called a domain of causality. The idea of splitting the set of servers is not new, but the published solutions either use vector clocks and broadcast [19] 14] or use matrix clocks but only reduce the message timestamp size [22][11]. Our solution reduces both the message timestamp size and the control information size on servers. In addition, the modularity of the architecture allows it to be adapted to any physical or logical topology. A domain of causality is a group of servers in which the causal order is respected. ....

L. Rodrigues and P. Verssimo. Causal Separators and Topological Timestamping: an Approach to Support Causal Multicast in Large Scale Systems, In Proceedings of the 15 th International Conference on Distributed Systems, May 1995.

....distribution of autonomous enterprise systems and the large number of hosts, could give rise to scalability issues. At the best of our knowledge, group toolkits have been used only in department (i.e. small scale) local systems. Only a few prototypes ( 5] 17] and some theoretical papers ( 6] [24]) address the problem of using a group toolkit over a large scale distributed system. On the other hand, IIOP based systems could face heterogeneity exploiting the high wrapping capabilities provided by CORBA. The main drawback resulting from their above the ORB design is performance: using ....

L. Rodrigues, P. Verissimo: "Causal Separators and Topological Timestamping: an Approach to Support Causal Multicast in Large-Scale Systems". Proceedings of the15th International Conference on Distributed Systems , 1995.

....to it. This abstraction yields simplified solutions to many fundamental problems in distributed computing, such as atomic snapshot [1] management of replicated data [4] and monitoring of distributed applications [15] Several protocols for implementing causal ordering have been proposed [4, 6, 17, 16]. These protocols mainly differ by the assumptions they make on the communication patterns, the topological structure of the underlying network, the amount of control information used to enforce causal ordering, and in how fast the protocol is in sending and delivering messages. For protocols that ....

....the discussion, and allows us to concentrate on the new things in our architecture and protocol. In a recent paper, Rodrigues and Verissimo describe an approach which is based on causal separators for reducing the amount of control information in systems that span several network domains [16]. Their approach, however, yields an architecture that is not hierarchical. Also, it does not reduce the amount of control informationused withinthe same subnet domain, even though current LANs can have as many as several hundreds of machines in the same subnet. Our solution does not assume ....

[Article contains additional citation context not shown here]

L. Rodrigues and P. Verissimo. Causal Separators and Topological Timestamping: an Approach to Support Causal Multicast in Large-Scale Systems. In Proc. of the 15th International Conference on Distributed Systems, May 1995.

....H p and to D q . When a message, m, is delivered at q 6= s m , m s timestamp, Hm , is added to H q . Additionally, m is added to H q and D q . Theorem 1: Rules 1 4 enforce message causal delivery (see appendix for proofs) Rules 1 4 above are enough to enforce causal delivery of messages (see [18] for a proof) However, this solution suffers from a serious drawback as, unless some measures are taken to garbage collect redundant elements, the causal histories continue to grow indefinitely. In the next paragraphs we present an extra set of rules that allow the garbage collection of the ....

L. Rodrigues and P. Verissimo. Causal separators and topological timestamping: an approach to support causal multicast in large-scale systems. Technical report, IST - INESC, Lisboa, Portugal, 1994.

....the unpropagated messages mechanism allows us to achieve the same benefits as lazy replication with more transparency for the client and more control for the server. Furthermore, with our approach compression mechanisms can be automatically and transparently applied by the communication system [3,17]. More recently, 21] presented a replicated RPC running on top of Amoeba in which point to point RPC was used to contact a designated replicated service coordinator. This is primarily due to the closed nature of Amoeba groups, but the authors also stress the transparency advantages inherent in ....

L. Rodrigues and P. Verissimo. Causal separators and topological timestamping: an approach to support causal multicast in large-scale systems. Technical report, IST - INESC, Lisboa, Portugal, 1994.

....H p and to D q . When a message, m, is delivered at q 6= s m , m s timestamp, Hm , is added to H q . Additionally, m is added to H q and D q . Theorem 1: Rules 1 4 enforce message causal delivery (see appendix for proofs) Rules 1 4 above are enough to enforce causal delivery of messages (see [18] for a proof) However, this solution suffers from a serious drawback as, unless some measures are taken to garbage collect redundant elements, the causal histories continue to grow indefinitely. In the next paragraphs we present an extra set of rules that allow the garbage collection of the ....

L. Rodrigues and P. Verissimo. Causal separators and topological timestamping: an approach to support causal multicast in large-scale systems. Technical report, IST - INESC, Lisboa, Portugal, 1994.

....the unpropagated messages mechanism allows us to achieve the same benefits as lazy replication with more transparency for the client and more control for the server. Furthermore, with our approach compression mechanisms can be automatically and transparently applied by the communication system [3,17]. More recently, 21] presented a replicated RPC running on top of Amoeba in which point to point RPC was used to contact a designated replicated service coordinator. This is primarily due to the closed nature of Amoeba groups, but the authors also stress the transparency advantages inherent in ....

L. Rodrigues and P. Verissimo. Causal separators and topological timestamping: an approach to support causal multicast in large-scale systems. Technical report, IST - INESC, Lisboa, Portugal, 1994.

No context found.

L. Rodrigues and P. Verissimo, "Causal separators and topological timestamping an approach to support causal multicast in large-scale systems", In the Proceedings of the 15 International Conference on Distributed Computing Systems, Vancouver, British Columbia, Canada, May 1995.

No context found.

L. Rodrigues and P. Verissimo. Causal separators and topological timestamping: an approach to support causal multicast in large-scale systems. In Proceedings of the 15th International Conference on Distributed Systems, May 1995. 12

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