P. M. Melliar-Smith, L. E. Moser and D. Agarwal, "Ring-based Ordering Protocols", in Proc. Int. Conf. on Information Engineering, (Singapore), pp.882-891, December 2-5 1991.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....ordering protocol constructed on a partial order on messages (like the Trans and Total protocols [37, 39, 40] an effective but computationally expensive solution. Alternatively, a total order on messages can be built directly without constructing a partial order first (like the Totem protocol [1, 3, 38]) Some solutions have been presented that do not make use of a total ordering protocol, like the protocol by Ricciardi and Birman [47] this protocol is based on the assumption that failure detectors rarely make mistakes. In those cases in which a correct processor p is falsely suspected by the ....

P. M. Melliar-Smith, L. E. Moser and D. Agarwal, "Ring-based Ordering Protocols", in Proc. Int. Conf. on Information Engineering, (Singapore), pp.882-891, December 2-5 1991.

....port number 16 bit UDP l ength 16 bit UDP checksum data (i f any) 0 15 16 31 8 bytes 16 bit source port number Background Diploma Thesis NTH, 1996 17 Figure 2.3 9 Fields used for computation of UDP checksum 2.4. The Totem System 2.4.1. Introduction The Totem System described in (P. M. Melliar Smith et al. December 1991), D. A. Agarwal et al. November 1995) Y. Amir et al. November 1995) D. A. Agarwal, December 1994) and (L.E. Moser et al. June 1995) is a group communication protocol, developed at the University of California, Santa Barbara. It supports fault tolerant applications in which distributed ....

....Solaris 2.3, operating over a 10 Mbit sec and 100 Mbit sec Ethernet. Work is being done to port the system to DEC alpha HP machines, and PCs running Windows NT. The following sections give a summary of the most important features and characteristics of the Totem system. For more details see (P. M. Melliar Smith et al. December 1991), D. A. Agarwal et al. November 1995) Y. Amir et al. November 1995) D. A. Agarwal, December 1994) L.E. Moser et al. June 1995) C. A. Lingley Papadopoulos, 1994) and Appendix E. 2.4.2. The Model The network used in the Totem system can be modeled as a single local area network on ....

P. M. Melliar Smith, L. E. Moser and D. A. Agarwal, "Ring-Based Ordering Protocols," Proceedings of the IEE International Conference on Information Engineering, Singapore (December 1991), pp. 882-891.

....after a membership change was detected and its members have been determined. It guarantees delivery of messages sent in the old configuration so that extended virtual synchrony is preserved. The basic idea behind the ordering algorithm is not original work of the author. It was published in [MMA91]. The membership state machine and the algorithm for achieving extended virtual synchrony are original work of the author. The first part of the membership algorithm is based on the Transis membership algorithm [ADKM92b] 34 5.2.1 Message Ordering The main principle of this algorithm is to ....

P. M. Melliar-Smith, L. E. Moser and D. A. Agarwal. Ring-based Ordering Protocols. In Proceedings of the International Conference on Information Engineering, pages 882-891, December 1991.

....such as non blocking atomic commitment protocols, total order broadcast multicast protocols. Usually, these protocols are considered separately, and do not rely on a common infrastructure. A notable exception is the group membership service [14] used to implement total order multicast protocols [2, 12, 5, 6]. However, the group membership problem (solved by the membership service) is just another example of an agreement problem that arises in distributed systems. All the agreement problems (atomic commitment, total order, membership) Research supported by the Fonds national suisse under contract ....

P. M. Melliar-Smith, L. E. Moser, and D. A. Agarwal. Ring-based ordering protocols. In Proceedings of the Int. Conference on Information Engineering, pages 882--891, December 1991.

....communication mechanism, processors may fail to receive some events. Processors must also maintain a consistent view of the set of processors involved in the computation so that they can track the departure and arrival of components involved in the computation. The Totem communication protocol [1, 7, 18] addresses these needs by providing reliable totally ordered message delivery across both local and wide area networks. Processors in a Totem network communicate by broadcasting messages. The protocol ensures that all messages broadcast in the network are eventually delivered to all destinations ....

P. M. Melliar-Smith, Louise E. Moser, and Deborah A. Agarwal. Ringbased ordering protocols. In Proceedings of the International Conference on Information Engineering, pages 882--891, Singapore, December 2-5 1991.

No context found.

Melliar-Smith, P. M., Moser, L. E., and Agarwal, D. A. Ring-based ordering protocols. In Proceedings of the IEE International Conference on Information Engineering (Singapore, December 1991), pp. 882--891.

....consists of models of the processors and of the network. The processor models execute the Totem protocol suite and use the model of the communication medium to multicast messages. 1. 1 Motivation The need for a protocol development environment arose during the development of the Totem system [1, 2, 6, 20, 22, 25] at the University of California, Santa Barbara. The initial protocol development environment for Totem [12, 13] provided a centralized simulator that allows a developer to monitor and control protocol execution for each processor in a single LAN. The user can single step protocol execution and ....

P. M. Melliar-Smith, L. E. Moser, and D. A. Agarwal, "Ring-based ordering protocols, " In Proceedings of the IEE International Conference on Information Engineering, pp. 882-891, Singapore, December 1991.

....an asynchronous event triggered approach cannot predetermine when messages will be generated, thereby precluding preplanned transmissions. To regulate the transmission of messages, such an approach often uses a token based protocol such as the Timed Token Protocol [3, 7] or the Totem Protocol [2, 8] which we analyze here. Such protocols typically depend on acknowledgment and retransmission mechanisms to overcome communication faults. The disadvantage of such protocols for real time systems is that they introduce two sources of uncertainty, the delay until the token arrives and delays caused ....

....must be updated, and those updates must be consistent even in the presence of faults and asynchrony, preferably without requiring difficult programming of the application. To meet the needs for fault tolerance in asynchronous real time distributed systems, we have developed the Totem Protocol [2, 8]. The Totem Protocol provides reliable totally ordered delivery of messages using a logical token passing ring imposed on a broadcast domain. The token contains ordering and retransmission information and is used to control access to the ring; only the processor in possession of the token can ....

P. M. Melliar-Smith, L. E. Moser and D. A. Agarwal, "Ring-based ordering protocols," Proceedings of the IEE International Conference on Information Engineering, Singapore, December 1991, pp. 882-891.

....[13, 15] to enable the user to observe processor interactions as they occur and also to trace interactions leading up to a particular event. These tools have proved to be extremely useful in ascertaining the causes and correctness of events. 2 The Totem Protocol The Totem single ring protocol [1, 4, 14] provides reliable totally ordered delivery of messages in a localarea network in which every message is broadcast to all of the processors. It is designed to operate in a distributed system, where fault tolerance is required, and to provide high performance. The underlying communication medium is ....

P. M. Melliar-Smith, L. E. Moser and D. A. Agarwal, "Ring-based ordering protocols," Proceedings of the IEE International Conference on Data Engineering, Singapore (December 1991), pp. 882-891.

....[8] We have focused on providing an additional level of knowledge within the group communication protocol, which allows us to reach agreement on a consistent transition from one configuration to another with minimal overhead. We show how our group communication techniques for Transis and Totem [2, 3, 18] can be utilized for database replication with high efficiency and strictly serializable semantics. 2 The Model 2.1 The Environment A replication service maintains an object in a distributed system. This object is replicated to improve performance and availability. The replication service is ....

....the need for separate agreements on the claiming of locks, since such agreements can be derived from the ordering of messages. Group communication can exploit hardware broadcasting or multicasting when it is available. Reliable ordered delivery protocols that use hardware broadcasts and multicasts [2, 3, 18, 19] can achieve higher throughput and lower latency than protocols based on point to point communication. If the system partitions into several components, the replication layer identifies at most one component as the primary component. The replication servers in a primary component determine the ....

[Article contains additional citation context not shown here]

P. M. Melliar-Smith, L. E. Moser, and D. A. Agarwal. Ring-based ordering protocols. In Proceedings of the IEE International Conference on Information Engineering, pages 882--891, December 1991.

....of the processors as well as of the network. The processor models execute the Totem protocol and use the model of the communication medium to pass messages. Motivation The need for a protocol development environment arose during the development of the Totem multi protocol system [1] 2] 4] 18][20][23] at the University of California, Santa Barbara. The initial protocol development environment for Totem [10] 11] provides a centralized simulator that allows a developer to monitor and control protocol execution for each processor in a single LAN. The user can single step protocol execution as ....

P. M. Melliar-Smith, L. E. Moser, and D. A. Agarwal, "Ring-Based Ordering Protocols." In Proceedings of the IEE International Conference on Information Engineering, pp. 882-891, Singapore, December 1991.

....Its novel mechanisms prevent delivery of messages in different orders in different components of a partitioned network, and provide accurate information about which processors have delivered which messages. Earlier versions of the Totem single ring protocol are described in [Amir et al. 1993; Melliar Smith et al. 1991]. Programming the application is considerably simplified if messages are delivered in total order rather than only in causal order, or if messages are delivered in causal order rather than only in FIFO order. In prior systems, delivery of messages in total order has been more expensive than ....

Melliar-Smith, P. M., Moser, L. E., and Agarwal, D. A. 1991. Ring-based ordering protocols. In Proceedings of the IEE International Conference on Information Engineering (Singapore). IEE, Stevenage, Herts, U.K., 882--891.

....fault tolerance and real time requirements. Unfortunately, reaching agreement in asynchronous dis 2 CHAPTER 1. INTRODUCTION tributed systems, when faults can occur, is non trivial. There exist strong impossibility proofs that it cannot be done deterministically [1] The Totem system [2, 3, 4, 5, 6, 7, 8, 9], developed at the University of California, Santa Barbara, was designed to address these issues [10] It makes the programming of distributed fault tolerant applications easier, faster and more robust. Programming of applications is much simpler if the messages are delivered in total order rather ....

P. M. Melliar-Smith, L. E. Moser and D. A. Agarwal, Ring-Based Ordering Protocols, Proceedings of the IEEE International Conference on Information Engineering, Singapore (December 1991), pp. 882-891.

....delivers messages to application processes in the appropriate process groups, and provides process group membership services. Detailed descriptions of the protocols that comprise the Totem system can be found in [2, 11] and descriptions of earlier versions of the protocols can be found in [1, 4, 13]. The Totem system has been implemented in the C programming language on Sun IPC workstations running SunOS 4.1, and on Sun SPARCstation 20s running Solaris 2.3, over a 10 Mbit sec Ethernet. It uses standard UNIX facilities, in particular UNIX UDP sockets to broadcast messages and to transfer the ....

P. M. Melliar-Smith, L. E. Moser, and D. A. Agarwal, "Ring-based ordering protocols," Proceedings of the IEE International Conference on Information Engineering, Singapore, pp. 882--891, December 1991.

....virtual synchrony which ensure that messages are not lost or incorrectly ordered as a result of reconfiguration. This work is based on prior experience with the Trans and Total protocols [13, 15, 16] which were effective but computationally expensive and led to the design of the Totem protocol [1, 12]. Ideas from the Trans protocol also influenced the development of the Transis system [2, 3] The work presented here extends the membership algorithm of Transis to handle ring formation and token generation and combines that algorithm with the single ring total ordering algorithm of Totem to ....

P. M. Melliar-Smith, L. E. Moser, and D. A. Agarwal. Ring-based ordering protocols. In Proceedings of the International Conference on Information Engineering, pages 882--891, Singapore, December 2-5 1991.

....most accurate available information, obtained from the sensor with the best view should be displayed to the operator. In the case of a network partition, however, it is better to display lower quality information from the connected sensors than to do nothing. In the design of the Totem protocol [3, 14], based on our experience with the Trans and Total protocols [13] and the Transis system [1, 2] we have extended the virtual synchrony model [4, 5, 6] of the Isis system to handle network partitioning and remerging and process failure and recovery. Extended virtual synchrony establishes a ....

P. M. Melliar-Smith, L. E. Moser and D. A. Agarwal, "Ring-based ordering protocols," Proceedings of the International Conference on Information Engineering, Singapore (December 1991), pp. 882--891.

....order of events seen by all processors in the execution of the distributed computation. A common order of events ensures identical executions and identical data at each processor, while different orders seen by two or more processors may result in inconsistencies in the data. The Totem protocol [1,2,12] addresses the need for consistency of replicated data by providing reliable totally ordered delivery of messages. The underlying communication medium on which the protocol executes is assumed to have an unreliable broadcast capability. Both processors and the communication medium may suffer from ....

....two or more rings are connected by gateways. Timestamps from approximately synchronized clocks are used to order messages consistently across rings, while sequence numbers on the individual rings are used to provide reliable delivery of messages. More details of the Totem protocol can be found in [1,2,12]. 2.2 Implementation of the Totem Protocol The implementation of the Totem protocol has been developed for the Unix operating system and consists of a set of portable software modules that provide reliable ordered delivery and processor membership services. Figure 2 presents a block diagram of ....

P. M. Melliar-Smith, L. E. Moser and D. A. Agarwal. Ring-based ordering protocols. In Proceedings of the International Conference on Information Engineering, pages 882-891, Singapore, December 1991.

....Computer scientists and communication engineers need proper tools for designing, developing, and debugging network protocols to produce reliable means of transporting data between sites in the system. One such protocol is Totem, developed at the University of California, Santa Barbara [1] 2] 5][21]. Totem is a fast and highly robust communication protocol for asynchronous distributed systems. Totem provides reliable ordered message delivery in a broadcast domain with processor membership, flow control, and recovery mechanisms which make it tolerant to faults such as message loss, site ....

....network. If the messages are received at all sites in the same order, we can be sure that processors will behave in exactly the same manner and replicated data will remain consistent. Different message ordering may result in inconsistent execution and data at each site. The Totem Protocol [1] 2] 5][21] addresses this need of consistency by providing reliable totally ordered delivery of messages multicast to groups of processors in a broadcast domain where all sites of the domain receive the message. Messages in a single broadcast domain are handled by the Totem Single Ring Protocol, and ....

P. M. Melliar-Smith, L. E. Moser, and D. A. Agarwal, "Ring-Based Ordering Protocols." In Proceedings of the International Conference on Information Engineering, pp. 882-891, Bangkok, Singapore, December 1991.

....meaning that the relative order of any two messages is determined directly from the messages, as broadcast by their sources. The Totem system uses born ordered messages, but some other multicast group communication systems do not. The Totem Single Ring Protocol The Totem single ring protocol [1, 3, 13] provides reliable totally ordered delivery of messages using a logical token passing ring superimposed on a LAN, such as an Ethernet. The token circulates around the ring as a point to point message, with a token retransmission mechanism to guard against token loss. Only the processor holding the ....

....the transitional configuration, the remaining messages of the old configuration are delivered. After these messages are delivered, the second Configuration Change message is delivered, introducing the new regular configuration. The Totem Multiple Ring Protocol The Totem multiple ring protocol [1, 2, 13] operates over multiple LANs interconnected by gateways. Imposed on each LAN is a logical token passing ring on which the singlering protocol operates. The multiple ring protocol provides essentially the same services, with the same properties, as the single ring protocol. In particular, the ....

Melliar-Smith, P. M., Moser, L. E., and Agarwal, D. A. Ring-based ordering protocols. In Proceedings of the IEE International Conference on Information Engineering (December 1991, Singapore) 882--891.

....and its relevance to this project in Section 3. In Section 4 we discuss the implementation and some of the issues involved. Section 5 deals with the relative advantages of this approach. Finally, we end with conclusions and a discussion of future work. 2 The Totem Protocol The Totem protocol [2, 3, 4, 10], which was used in this project, is a fault tolerant communication protocol that provides reliable ordered delivery of messages. The Totem protocol was originally designed to run over a single localarea network with a logical token passing ring superimposed on it. Now it has been extended to run ....

P. M. Melliar-Smith, L. E. Moser, and D. A. Agarwal. Ring-based ordering protocols. In Proceedings of the International Conference on Information Engineering, pages 882--891, Singapore, December 1991.

....most accurate available information, obtained from the sensor with the best view should be displayed to the operator. In the case of a network partition, however, it is better to display lower quality information from the connected sensors than to do nothing. In the design of the Totem protocol [3, 12], based on our experience with the Trans and Total protocols [11] and the Transis system [1, 2] we have extended the virtual synchrony model [4, 5, 6] of the Isis system to handle network partitioning and remerging, as well as process failure and recovery. Extended virtual synchrony establishes a ....

P. M. Melliar-Smith, L. E. Moser and D. A. Agarwal, "Ring-based ordering protocols," Proceedings of the International Conference on Information Engineering, Singapore (December 1991), pp. 882--891.

....network is a node and each gateway is an edge, then network connectivity can be analyzed as a graph partitioning problem. We have investigated graphs of networks that were constructed by adding random edges to the graph and robust graphs that were constructed to be resilient to edge deletion [41]. The data indicate that, as gateways fail, partitioning of the network is of more concern than the increased length of the routes in the network. To provide efficient broadcast communication, effective flow control is required. Broadcast communication can proceed only at the rate of the slowest ....

....despite the loss of contact with a subset of the nodes. The robust graphs, however, remained connected on average five to seven edge deletions longer than the random graphs with the smaller disconnected component being only one or two nodes in over 99 of the cases. This work is also reported in [41]. 5.5 Proof of Correctness Membership Uniqueness of Topologies Theorem 5.1 Each topology identifier is unique; moreover, at any time a processor or gateway is a member of at most one topology. Proof. A topology identifier is the lexicographically ordered list of ring identifiers of the rings ....

P. M. Melliar-Smith, L. E. Moser, and D. A. Agarwal. Ring-based ordering protocols. In Proceedings of the International Conference on Information Engineering, pages 882--891, Singapore, December 2-5 1991.

....delivers messages to those application processes that are members of the appropriate process group, and maintains the process group membership. In this paper we do not discuss the Totem singlering protocol or the physical layer below it; descriptions of the single ring protocol can be found in [1, 3, 10]. Rather, we provide a high level description of the Totem multiple ring protocol. A detailed description of the multiple ring protocol, including proofs of correctness and pseudo code for the implemented protocol, can be found in [1] The Totem multiple ring protocol is the first protocol of ....

....Preacknowledgments are used to ensure that buffers do not overflow and to define a total order on messages. The RMP protocol [17] also provides reliable ordered multicasts over larger networks, but lacks the performance of UDP over a single local area network. The Totem single ring protocol [1, 3, 10] was designed to achieve lower computational cost and higher throughput than the Trans and Total protocols. It provides totally ordered message delivery and membership services using a logical token passing ring imposed on a local area network with broadcast communication. Consistency of message ....

P. M. Melliar-Smith, L. E. Moser, and D. A. Agarwal. Ring-based ordering protocols. In Proceedings of the IEE International Conference on Information Engineering, pages 882--891, Singapore, December 1991.

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