Yair Amir, Danny Dolev, Shlomo Kramer, Dalia Malki. Membership Algorithms in Broadcast Domains. In Proceedings of the 6th WDAG; (Isreal, 1992). Springer Verlag Lecture Notes in Computer Science 647, 292312.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....operation basis, depending on the needs of the application and the costs they are prepared to pay for this guarantee. 13.10 The Group Membership Problem The role of a group membership service (GMS) is to maintain the membership of a distributed system on behalf of the processes that compose it [BJ87b, Cri91b, MPS91, MSMA91, RB91, ADKM92b, Gol92, Ric92, Ric93, RVR93, Mal94, MMA94, Aga94, BDGB94, Rei94b BG95, CS95, ACBM95, BDM95, FKMBD95, CHTC96, GS96]. As described above, processes join and leave the system dynamically over its lifetime. We will adopt a model in which processes wishing to join do so by first contacting the GMS, which updates the list of system members and then grants the request. Once admitted to the system, a process may ....

....component of the GMS. We will use the term view of the GMS membership to denote the value of the membership list that holds for a given process within the GMS at a specified point in its execution. If the GMS can experience a partitioning failure, it can also experience the merging of partitions [ADKM92b, Mal94, MMA94]. The GMP should therefore include a merge protocol. Finally, if all the members of the GMS fail, or if the primary partition is somehow lost, the GMP should provide for a restart from complete failure, or for identification of the primary partition when the merge of two nonprimary partitions ....

[Article contains additional citation context not shown here]

Yair Amir, Danny Dolev, Shlomo Kramer, Dalia Malki. Membership Algorithms in Broadcast Domains. In Proceedings of the 6th WDAG; (Isreal, 1992). Springer Verlag Lecture Notes in Computer Science 647, 292312.

....RHODOS November 28, 1997 Page 1 1 Introduction The group communication paradigm provides a powerful operating system abstraction for the development of distributed applications and services. However, until recently only a few operating systems provide the basic support for this abstraction [Amir et al. 1992a] Tanenbaum 1990] Group communication can be used in fault tolerant systems for the propagating updates to replicas, in parallel systems to distribute jobs among slave worker processes, and in distributed systems to support parallel execution on a set of workstations, and for applications such ....

....does not attempt to provided any fault tolerance instead allowing the application layer programmer to perform the required reliability measure for correct operation. 6. 5 Transis Transis is an efficient and reliable transport protocol currently under development at Hebrew University, Jerusalem [Amir et al. 1992a] Transis has been developed to provide support for a wide variety of reliable multicast message passing services between processors for UNIX. Transis operates within a communication broadcast domain where processors can initiate multicast messages to other processor within a chosen subset of the ....

Y. Amir, D Dolev, S. Kramer, and D. Malki., "Membership Algorithms in Broadcast Domains," Computer Science Technical Report, CS 92-10, Computer Science Department, The Hebrew University of Jerusalem, Israel.

....can be un reachable for some period of time. This is further complicated as in an asynchronous environment, such as a distributed system, it is almost impossible to distinguish The Client Server Model and Systems 14 between the slow response of a remote process to that of a partitioned network [Amir et al. 1992]. Hence, the group membership facility must provide the flexibility to maintain serviceable performance during network partitioning (although slightly degraded) by allowing a subgroup to form inside the communicating group. When the network partitions disappear the group membership support should ....

Y. Amir, D Dolev, S. Kramer, and D. Malki., "Membership Algorithms in Broadcast Domains," Computer Science Technical Report, CS 92-10, Computer Science Department, The Hebrew University of Jerusalem, Israel.

....pv i 1 is installed, v i 1;2;p and v i 1;2;q may also be denoted by v i 1; p and v i 1; q , respectively. There are a few other minor differences (refer to [7] for more details) Some virtual synchrony protocols allow multiple groups to exist at the same time (multiple view membership protocols)[1, 2, 3, 7], whereas others allow at most one group to exist at any time (singleview membership protocols) 4, 12] Our protocol is a single view membership protocol. To enforce this, when installing permanent view pv i 1 , a majority of nodes in pv i must be alive; otherwise, the whole system halts. Note ....

Y. Amir, D. Dolev, S. Kramer, and D.Malki. Membership algorithms in broadcast domains. In Preceedings of the 6th International Workshop on Distributed Algorithms, Lecture Note in Commputer Science 647, pages 292--312, 1992.

....of process group execution models, including the virtual synchrony model first introduced by the Isis toolkit. This model was also adopted by [2] and [16] Horus can be configured to allow progress during transient failure and network partitions, using a variation of protocols proposed by Transis [1]. When a network partition failure occurs, a single group may split into multiple subgroups: one primary and others non primary subgroups. Group members in different subgroups will then observe different sequences of views. When the partition is repaired a non primary subgroup can heal itself by ....

Yair Amir, Danny Dolev, Shlomo Kramer, and Dalia Malki. Membership algorithms in broadcast domains. In A. Segall and S. Zaks, editors, Proceedings of the Sixth WDAG; Israel, pages 292--312. Springer-Verlag, 1992.

....be implemented using independent RMP groups. However, this approach is inefficient because each group independently performs the same type of group management services. RMP provides no optimization for sharing resources (name service, fault detection, etc. among related groups. Transis and Totem [2, 18] extend virtual synchrony to tolerate network partitioning. Like ISIS, these systems are designed for applications that require high reliability. The existing system offers limited communication services to support collaborative and multimedia applications. A recent proposal [6] called Multimedia ....

....although the delay is palpable. We hope to incorporate additional application driven optimizations such as compression in the coming months to improve performance for this class of applications. 7 Comparison with other group systems In this section, we compare CCTL to Horus [27, 10] Transis [2, 6] and RMP [28] and show how to build CCTL using a Horus protocol stack. 7.1 Multiple service options in Horus Horus group protocols can be dynamically composed by stacking sub protocol layers such as: COM, NAK, MBRSHIP, STABLE, WVS and TOTAL. COM interfaces to standard network protocols such as ....

Y. Amir, D. Dolev, S. Kramer, and D. Malki, "Membership algorithms in broadcast domains", 6th Int. Workshop on Distributed Algorithms, Lecture Notes in Computer Science #674, pp. 292--312, Nov. 1992.

....time of a light weight group leave event for both the local and remote view manager cases. Under both situations the cost of a light weight group leave was 9 ms. The cost of a core group leave for the remote case was 197 ms, and for the local leave it was 80 ms. 8 Related Work The Transis system [3, 2] provides process sets at the session layer of their system. These are closely related to the multiplexing layer of HORUS. Their job is to map a process abstraction of membership onto a site abstraction of membership. Transis has a single Lansis process on each processor which coordinates the ....

Yair Amir, Danny Dolev, Shlomo Kramer, and Dalia Malki. Membership algorithms in broadcast domains. Technical Report CS92-10, The Hebrew University of Jerusalem, June 1992.

....resources (name service, fault detection, etc. among related groups. In addition, RMP provides only virtually synchronous atomic multicast. Applications requiring only reliable FIFO message ordering with virtual synchrony are forced to use the more expensive total ordering. Transis and Totem [2, 14] extend virtual synchrony to tolerate network partitioning. The existing system offers limited service qualities although a recent proposal [6] provides additional flexibility. Reliable messages are used in a novel way to coordinate and synchronize streams offering differing service qualities. ....

....delivers fail messages for q to C and resumes message processing. The optimized protocol requires no additional message retransmission and has the same message complexity as the fault free protocol. 5 Comparison with other systems In this section, we compare CCTL to Horus [23, 9] Transis [2, 6] and RMP [24] and show how to build CCTL using a Horus protocol stack. 5.1 Multiple service options in Horus Horus group protocols can be dynamically composed by stacking sub protocol layers such as: COM, NAK, MBRSHIP, STABLE, WVS and TOTAL. COM interfaces to standard network protocols such as ....

Y. Amir, D. Dolev, S. Kramer, and D. Malki, "Membership algorithms in broadcast domains", 6th Int. Workshop on Distributed Algorithms, Lecture Notes in Computer Science #674, pp. 292--312, Nov. 1992.

....systems in [Cri91b] Since then this problem has also been subject of intense investigation for asynchronous systems. Two types of asynchronous membership services have been described: primary partition membership [RB91, KT91, MPS92, MSPMA94, HS95, MFSW95, Nei96] and partitionable membership [ADKM92, JFR93, vRBM96, BBDG94, EMS95]. Some of these membership services explicitly assume a time free underlying system model (e.g. RB91, MPS92, HS95, BBDG94] while for others it is not entirely clear whether the underlying system model is time free or some variant of our timed system model. The primary partition services are ....

Y. Amir, D. Dolev, S. Kramer, and D. Malki. Membership Algorithms in Broadcast Domains. Technical Report CS92-10, Computer Science Department, The Hebrew University of Jerusalem, Israel, 1992.

....such system, it is appropriate for the membership service to deliver views (even concurrent ones) to all functional components and let the application decide if progress is possible. The resulting service is called partial group membership in that the set of installed views defines a partial order [1, 21, 27, 6]. The final issue has to do with the composition of concurrent views. If two concurrent views are allowed to overlap arbitrarily in their composition, the service is called weak partial group membership [21, 27] If, on the other hand, concurrent views are guaranteed to be disjoint in their ....

Y. Amir, D. Dolev, S. Kramer and D. Malki. Membership Algorithms in Broadcast Domains. In Proc. 6th Intl. Workshop on Distributed Algorithms, A. Segall and S. Zacks (Eds.), Haifa, Israel, Lecture Notes in Computer Science, vol. 647, Springer-Verlag, November 1992, 292--312.

....communication : 1) at the lowest level, the FS component; on top of it, 2a) a component that defines new views, and 2b) a component that reliably multicasts messages within a view. 1 Introduction There have recently been several papers about membership services in asynchronous systems [2, 12, 16, 17, 18]. A membership service is responsible for giving each process (consistent) information about the operational processes in the system. A process calls this information its view of the system processes. A membership service typically reacts to process crashes or recoveries, leading it to define a ....

....in which processes may crash and the network may partition. However, despite network partitions, this membership service defines only majority views a unique, totally ordered sequence of views. Such a membership service is said to have linear semantics. ffl The membership services described in [1, 2, 12] consider the same failure scenario as above, but only define a partial order on the views. That is, if the system is partitioned in two (or more) subnetworks then two (or Paper appeared in IEEE Proc of the 23rd Annual Int. Symp. on Fault Tolerant Computing (FTCS23) Toulouse, June 22 24, ....

[Article contains additional citation context not shown here]

Y. Amir, D. Dolev, S. Kramer, and D. Malki. Membership Algorithms in Broadcast Domains. In A. Segall and S. Zaks, editors, Proceedings of the Sixth WDAG; Israel, pages 292--312. Springer-Verlag, 1992. LNCS 647.

....communication channels and makes an efficient use of the network broadcast capability. A full description of all of the Transis services is beyond the scope of this paper. In this paper, we give an overview of the system structure and its basic protocols. The interested reader is referred to [1, 2] for more details. Melliar Smith et al. suggest in [16, 17] a novel protocol for reliable broadcast communication over physical LANs, the Trans protocol. Similar ideas appear in the Psync protocol ( 19] These protocols use the hardware broadcast capability for message dissemination and a ....

....Consequently, the membership algorithm never allows blocking, and operates within the regular flow of messages. The sections bellow give the essentials of the algorithm and an intuitive claim of its correctness. A full description of the membership algorithm and its proof is provided in [1]. 5.1 The Faults Handling Algorithm This section focuses on a membership algorithm for handling departure of processors from the set of active ones. Throughout this section, we assume the existence of a starting current membership, Members, which is the agreed set of connected members. ....

Y. Amir, D. Dolev, S. Kramer, and D. Malki. Membership algorithms in broadcast domains. Technical Report CS92-10, dept. of comp. sci.., the Hebrew University of Jerusalem, 1992.

No context found.

Amir, Y., Dolev, D., Kramer, S., and Malki, D. Membership algorithms in broadcast domains. In Proceedings of the 6th International Workshop on Distributed Algorithms (Haifa, Israel, November 1992), pp. 292--312. The Totem Single-Ring Ordering and Membership Protocol \Delta 29

No context found.

Amir, Y., Dolev, D., Kramer, S., and Malki, D. 1992a. Membership algorithms in broadcast domains. In Proceedings of the 6th International Workshop on Distributed Algorithms (Haifa, Israel). Springer-Verlag, Berlin, Germany, 292--312.

....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 obtain a highly efficient fault tolerant total ordering protocol. 2 Related Work An early ....

Y. Amir, D. Dolev, S. Kramer, and D. Malki. Membership algorithms in broadcast domains. In Proceedings of the 6th International Workshop on Distributed Algorithms, Lecture Notes in Computer Science 647, pages 292--312, Haifa, Israel, November 1992.

....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 consistent relationship between delivery of messages and delivery of configuration changes across all ....

Y. Amir, D. Dolev, S. Kramer and D. Malki, "Membership algorithms in broadcast domains," Proceedings of the 6th International Workshop on Distributed Algorithms, Haifa, Israel (November 1992), Lecture Notes in Computer Science 647, pp. 292-312.

....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 consistent relationship between delivery of messages and delivery of configuration changes ....

Y. Amir, D. Dolev, S. Kramer and D. Malki, "Membership algorithms in broadcast domains," Proceedings of the 6th International Workshop on Distributed Algorithms, Haifa, Israel (November 1992), Lecture Notes in Computer Science 647, pp. 292-312.

....of the active processors. The basis layer of Lansis, membership, automatically maintains the CCS in consensus among all the members of CCS. The Membership section below sheds some intuition on the membership algorithm. The full details of the algorithm and proof of its correctness are found in [1]. The next layer on top of the dynamic membership is responsible for reliable delivery of messages within the CCS. It delivers both regular messages and special configuration change messages to the upper level. Lansis supports various primitives for coordinating the delivery of multicast messages ....

Y. Amir, D. Dolev, S. Kramer, and D. Malki. Membership algorithms in broadcast domains. Technical Report CS92-10, dept. of comp. sci.., the Hebrew University of Jerusalem, 1992.

No context found.

Y. Amir, D. Dolev, S. Kramer, and D.Malki. Membership algorithms in broadcast domains. In Preceedings of the 6th International Workshop on Distributed Algorithms, Lecture Note in Commputer Science 647, pages 292--312, 1992.

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