Eric C. Cooper. Replicated procedure call. In Proceedings of the 3rd ACM symposyum on Principles of Distributed Computing, Berkeley, CA 94720, USA, August 1984. ACM.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....is called a server. For fault tolerance, the ambassador may be replicated. In this case, two or more ambassadors will serve remote invocations, and each ambassador will keep a local image of the object. Clients will interact with the group of ambassadors, through replicated remote invocations 1 [6,11] (see figure 3) 1 When objects are accessed exclusively through remote invo Ambassador Group Remote Client Group (1 element only) Data Data Node C Node D Node B Node A C1 C2 CR2 CR1 CR2 CR1 GRIP Matching Ambassador Client Client Repres. Node Data Copy Data CR1 C1 Figure ....

....remote invocations and local copy access. To our knowledge, the work in this area is very limited. However, each of these methods have been intensively studied in separate. Remote invocations (and remote function calls) to replicated objects has been supported by several systems as troupes [6], ARGUS [11] or DELTA 4 [15] More recent research projects include the ELECTRA [12] object oriented toolkit that also provides abstractions for remote method calling or the Object Groups [10] prototype that uses the ISIS system. However, most of these systems preclude local access to replicas ....

Eric C. Cooper. Replicated procedure call. In Proceedings of the 3rd ACM symposyum on Principles of Distributed Computing, Berkeley, CA 94720, USA, August 1984. ACM.

....allows the client process to synchronize with the results of such parallel calls. The first version of the Nexus RPC facility, which we implemented in 1988, supported this concept of asynchronous calls [14] Several other RPC systems have also supported asynchronous remote procedure calls [30] [31] 32] 33] and have demonstrated the benefits of such as a facility. A survey of various asynchronous RPC facilities is presented in reference [34] most of these systems were developed in periods overlapping or following the development of the Nexus RPC system. Some of these systems are ....

Eric Cooper, `Replicated Procedure Call', Proceedings of the Third Annual Symposium on Principles of Distributed Computing, 220--232 (1984).

....has been a research issue for more than two decades. Although often limited in earlier systems, support for fault tolerance and replication has been subject to a growing interest in the last few years. Several approaches have been proposed, including many in the area of remote invocation [7,9,8]. However, most existing systems provide support for at most a limited set of pre defined replication strategies. Approaches like problem oriented shared memory [6,16,11] which allow different replication strategies to be applied to different objects are a key factor for efficiency in distributed ....

....existing approaches are designed with a specific replication scheme and thus are not concerned with some of the problems that our protocols tackle. Here we relate our work to some of the more relevant examples of replicated invocation protocols. One of the first examples was presented by Cooper in [7]. The protocol allows invocation of a troupe of replicas, but is more restrictive than ours; it requires programs to be completely deterministic and it assumes that all troupe members receive and process all requests. ISIS[3] offers a multicast communication system that supports virtual synchrony ....

Eric C. Cooper. Replicated procedure call. In Proceedings of the 3rd ACM symposyum on Principles of Distributed Computing, Berkeley, CA 94720, USA, August 1984. ACM.

....assumption (both timing and value errors are detected) Any detected errors cause the incriminated endpoint replica to be aborted and are reported to the administration system to initiate fault treatment. Active replication has also been studied (under various names) in the SIFT[13] CIRCUS[6], ISIS[4] and AAS[8] projects. To our knowledge, however, the Delta 4 implementation is the only one that manages replicated output messages at their logical source rather than at their destination and allows arbitrary (host) failures to be tolerated over a general (non meshed) 8 communication ....

....knowledge, however, the Delta 4 implementation is the only one that manages replicated output messages at their logical source rather than at their destination and allows arbitrary (host) failures to be tolerated over a general (non meshed) 8 communication network. To be fair, although Cooper[6] assumed that processors only failed by crashing, he did study the concept of a collator to vote on multiple replies from nondeterministic server replicas. Passive replication is a technique in which only one of the replicas (the primary replica) processes the input messages and provides ....

E. C. Cooper, "Replicated Procedure Call", ACM Op. Sys. Review, 20 (1), pp.44-56, 1984.

....server troupe model. Multiple process rendezvous [1] enable synchronization of an arbitrary number of processes. MultiRPC [16] is an incorporation of multicast communication into an RPC based distributed sequential programming. Another method of group communication based on RPCs is Replicated RPC [8] aimed at programming replicated services. A novel approach to specification of object oriented reactive systems based on multi object actions has also been devised [10] Fragmented Objects (FO) 13] are aimed at supporting abstractions of distribution and provide a library of objects which may be ....

Eric C. Cooper. Replicated Procedure Call. Operating Systems Review, 20(1):44--56, January 1986.

....replicas of that process continue the computation without interruption. The methods of N modular redundancy and N version programming are special cases of this class of fault tolerance methods. Examples of systems using active replication include the new ISIS system (ISIS 2 ) Birman87] Circus [Cooper85, Cooper84], CHORUS [Banino85, Banino82] MP [Gait85] MARS [Kopetz85b] FT Concurrent C [Cmelik88] PRIME [Fabry73] SIFT [Wensley78] and Yang and York s work on the Intel iAPX 432 [Yang85] These methods are well suited for use in real time systems, since failure recovery is essentially immediate. ....

Eric C. Cooper. Replicated procedure call. In Proceedings of the Third Annual ACM Symposium on Principles of Distributed Computing, pages 220--232. ACM, August 1984.

....replicated objects. For fault tolerance, the ambassador may be replicated. In this case, two or more ambassadors will serve remote invocations, and each ambassador will keep a local image of the object. Clients will interact with the group of ambassadors, through replicated remote invocations [9,21] (see figure 3) 1 . 2.2 Maintain a local image Another method to access ROMANCE objects is to maintain a local image of the object. This may improve performance if locality of data can be exploited. However, to use this approach, one has to solve twoproblems: i) when the new image is created, ....

....remote invocations and local copy access. To our knowledge, the work in this area is very limited. However, each of these methods have been intensively studied in separate. Remote invocations (and remote function calls) to replicated objects hasbeen supported by several systemsas troupes [9], ARGUS [21] DELTA 4 [27] or Arjuna [31] More recent research projects include the ELECTRA [24] object oriented toolkit that also provides abstractions for remote method calling or the Object Groups [18] prototype that uses the ISIS system. However, most of these systems preclude local access to ....

Eric C. Cooper. Replicated procedure call. In Proceedings of the 3rd ACM symposyum on Principles of Distributed Computing, Berkeley, CA 94720, USA, August 1984. ACM.

....for process groups. The ISIS reliable distributed toolkit [2] also based on process groups and multicast, enables group communication to increase fault tolerance. Remote Procedure Call has been enhanced to handle communication with multiple processes simultaneously [16] and to support replication [6]. The success of group communication in distributed systems provoked research into adopting the concept as a programming technique. Broadcasting Sequential Processes [7] enabled sequential processes similar to those in Hoare s CSP [9] to broadcast messages to all others although there was no ....

E. C. Cooper. Replicated Procedure Call. Operating Systems Review, 20(1):44--56, January 1986.

....early systems, the support for fault tolerance and replication has been subject to a growing interest in the last few years. Several solutions have been proposed to tackle this problem including: support for replicated objects through remote invocations (and remote function calls) such as troupes [9], ARGUS [18] or DELTA 4 [26] language approach (such as ORCA [2] etc. However, most of these systems provide support for only a limited set of pre defined replication strategies. The use of a problem oriented shared memory approach [7] by allowing different replication methods to be applied ....

Eric C. Cooper. Replicatedprocedure call. In Proceedings of the 3rd ACM symposyum on Principles of Distributed Computing, Berkeley, CA 94720, USA, August 1984. ACM.

....in that recovery points are specified by the application. The semantics of these programming constructs specify a small subset of recoverable events. Overhead can be limited during failure free execution to those regions surrounding these events. Reliable RPC mechanisms such as those in [12] and [2] also specify recovery regions and use type information to further optimize rollback of modified state. The work described by Xu in [49] implements a persistent tuple space through an active replication scheme in which replicas need only be synchronized during out( operations, and then ....

Cooper, E. C. Replicated Procedure Call. in: ACM Symposium on Principles of Distributed Computing. 1984, pp. 220--232.

....offer coherence at a certain level of abstraction. The ActorSpace coordination primitives we will develop include broadcasting messages to groups of receivers, also known as multicasting. Broadcasting has been studied for some time now. Initial work was done on extending RPC to support replication [22], on broadcasting as a programming paradigm [25] and on protocols for reliable broadcasting [18] Later work has focused on protocol design [30, 34] and on improving and supporting reliability of broadcasting protocols [24, 13] In operating systems research, the notion of process groups has been ....

.... has also been discussed as a way of achieving fault tolerance in operating systems [9] Broadcasting provides a flexible way of replicating a service, where a client broadcasts a request to a group of servers, and collects one or more responses using 2 Related Work 6 some voting criteria [22]. ActorSpace builds on a concurrent object oriented programming paradigm (COOP) COOP systems, such as Actors [1] Emerald [29] Orca [10] and Concurrent Aggregates [20] support an object based programming model, where objects may invoke methods in other objects by giving a reference to the object ....

Eric C. Cooper. Replicated Procedure Call. In Proceedings of the Third Annual ACM Symposium on Principles of Distributed Computing Conference, pages 220--232. ACM, August 1984.

....the large spectrum of application requirements and the heterogeneity of interconnecting links, characteristics which are also shared by mobile systems. Unfortunately, in most systems a replicated service can be accessed remotely only via protocols tailored to its specific replication protocol [8, 19], causing an undesirable violation of both transparency and modularity. In order to maintain the transparency of an object replication scheme, we have designed a fault tolerant Generic Remote Invocation Protocol (GRIP) 24] which is independent of the service replication protocol and which ....

E. Cooper. Replicated procedure call. In Proceedings of the 3rd ACM symposyum on Principles of Distributed Computing, Berkeley, U.S.A., Aug. 1984.

....has been a research issue for more than two decades. Although often limited in earlier systems, support for fault tolerance and replication has been subject to a growing interest in the last few years. Several approaches have been proposed, including many in the area of remote invocation [7,9,8]. However, most existing systems provide support for at most a limited set of pre defined replication strategies. Approaches like problem oriented shared memory [6,16,11] which allow different replication strategies to be applied to different objects are a key factor for efficiency in distributed ....

....existing approaches are designed with a specific replication scheme and thus are not concerned with some of the problems that our protocols tackle. Here we relate our work to some of the more relevant examples of replicated invocation protocols. One of the first examples was presented by Cooper in [7]. The protocol allows invocation of a troupe of replicas, but is more restrictive than ours; it requires programs to be completely deterministic and it assumes that all troupe members receive and process all requests. ISIS[3] offers a multicast communication system that supports virtual synchrony ....

Eric C. Cooper. Replicated procedure call. In Proceedings of the 3rd ACM symposyum on Principles of Distributed Computing, Berkeley, CA 94720, USA, August 1984. ACM.

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