G. Morgan, S. K. Shrivastava, P. D. Ezhilchelvan and M.C. Little, "Design and Implementation of a CORBA Fault-tolerant Object Group Service", Proceedings of 2 nd IFIP International Working Conference on Distributed Applications and Interoperable Systems, DAIS'99, Helsinki, June 1999.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....in a partially synchronous 35 system, as asynchrony can cause the problem of system instability at the group communication level. The previous problem has even a greater impact on systems following the service approach, i.e. the Object Group Service (OGS) 49] and the NewTOP object group service [50] that build a group communication toolkit exploiting the ORB as the underlying communication channel and offer developers IDL interfaces to use such services. This implies that these toolkits are fully 40 portable and interoperable, but the developer must explicitly handle replica consistency ....

Morgan G, Shrivastava SK, Ezhilchelvan PD, Little MC. Design and implementation of a CORBA fault-tolerant object group service. Proceedings of the 2nd IFIP WG 6.1 International Working Conference on Distributed Applications and Interoperable Systems (DAIS'99), 1999; 179--185.

....for the transfer of application level state. Developed at the University of Newcastle, Newtop is a group communication toolkit that is exploited to provide fault tolerance to CORBA using the service approach. While the fundamental ideas are similar to OGS, the Newtop based object group service [8] has some key differences. Of particular interest is the way this service handles failures due to partitioning support is provided for a group of replicas to be partitioned into multiple sub groups, with each sub group being connected within itself. No mechanisms are provided, however, to ....

G. Morgan, S. Shrivastava, P. Ezhilchelvan, and M. Little. Design and implementation of a CORBA fault-tolerant object group service. In Proceedings of the Second IFIP WG 6.1 International Working Conference on Distributed Applications and Interoperable Systems, Helsinki, Finland, June 1999.

....for duplicate detection and suppression, and for the transfer of application level state. Newtop is a group communication toolkit that is exploited to provide fault tolerance to CORBA using the service approach. While the fundamental ideas are similar to OGS, the Newtop based object group service [6] has some key differences. Of particular interest is the way this service handles failures due to partitioning. Support is provided for a group of replicas to be partitioned into multiple sub groups, with each sub group being connected within itself. No mechanisms are provided to ensure consistent ....

G. Morgan, S. Shrivastava, P. Ezhilchelvan, and M. Little. Design and implementation of a CORBA fault-tolerant object group service. In Proceedings of the Second IFIP WG 6.1 International Working Conference on Distributed Applications and Interoperable Systems, Helsinki, Finland, June 1999.

....replica is approximately 2 seconds, most of which stems from the view agreement protocol. 6 Related Work Many other projects have also addressed the problems related to making distributed applications fault tolerant, most of which are based on group communication [8] Primarily, research efforts [13, 19, 12, 7, 18, 6, 21] have focused on two slightly different distributed object technologies, namely CORBA [23] and Java RMI [26] Unfortunately, most efforts have been dedicated to CORBA and resulting object group systems may be classified into three categories [12] The integration approach involves modifying and ....

....approach. Finally, the service approach provides group communication as a separate CORBA service. The ORB is unaware of groups, and the service can be used in any CORBA compliant implementation. The service approach has been adopted by object group systems such as OGS [12] DOORS [10] and Newtop [18]. Unlike CORBA, the specification of Java RMI enables programmers to implement their own remote references, thus extending the standard behavior of Java RMI. We have exploited this feature by developing the concept of group reference, whose task is to manage interactions between clients and ....

[Article contains additional citation context not shown here]

G. Morgan, S. Shrivastava, P. Ezhilchelvan, and M. Little. Design and Implementation of a CORBA FaultTolerant Object Group Service. In Proc. of the 2nd IFIP Int. Conf. on Distributed Applications and Interoperable Systems, pages 361--374, Helsinki, Finland, June 1999.

....categories. The rst approach is to create a fault tolerant ORB. Both Piranha [27] 29] and Maestro, fall into this category. The second category is involves proving fault tolerance through a CORBA service, above the CORBA Object Request Broker (ORB) The OpenDREAMS [30] project and Arjuna [31] [32] take this approach. 7 A third method is to intercept messages from the ORB; this is the approach taken by Eternal [33] 36] More speci cally, the rst approach consists of building an ORB that has built in fault tolerance capabilities. For example, Electra uses adapter objects to ....

G. Morgan, S. K. Shrivastava, P. D. Ezhilchelvan, and M.C. Little, \Design and implementation of a CORBA fault-tolerant object group service," in Proceedings of the Second IFIP WG 6.1 International Working Conference on Distributed Applications and Interoperable Systems, DAIS99, Helsinki, June 1999, pp. 177-195.

....IIOP messages. Eternal Immune [Nar97, Nar99] and AQuA apply this interception approach. The third approach involves the addition of group communication and fault tolerance mechanisms as CORBA services. This service approach is used by the OpenDREAMS project [Fel96] the Arjuna group [Mor99], and the DOORS project [Gok00] More specifically, Electra [Maf95] provides fault tolerance to CORBA by building a specialized ORB. The Electra ORB adds several properties of group communication systems to a common ORB. In particular, Electra allows dynamic replication of important object ....

....to build the group framework are group multicast, dynamic group membership, view change, and state transfer. This approach has the potential to provide group services to CORBA objects; however, it requires that the application developers be aware of, and explicitly make use of, the OGS. Note that [Mor99] uses the same approach, implementing a CORBA object group service by using the Newtop [Ezh95] group communication system. The DOORS [Gok00] project builds a fault tolerant CORBA service that provides active and passive replication but does not depend on reliable group communication. Several ....

G. Morgan, S. K. Shrivastava, P. D. Ezhilchelvan, and M. C. Little, "Design and Implementation of a CORBA Fault-tolerant Object Group Service," Proc. of the Second IFIP WG 6.1 International Working Conference on Distributed Applications and Interoperable Systems, DAIS'99, Helsinki, June 1999.

....for different # values (b) Scale out for different ## values (c) Scale out for ## ###and different # Figure 1: Scale out for different values of # and ## 2.2 Related Work Previous work on reliable middleware has mainly concentrated on replicated ORBs. Examples of these approaches are [Maf95, MSEL99, NMMS99, FGS99] In contrast, this paper focuses on a novel middleware layer for database replication. The limitations of replication pointed out in the formal model of the previous section are well known 5 among database designers. Gray et al. GHOS96] provided an empirical evaluation of them ....

G. Morgan, S.K. Shrivastava, P.D. Ezhilchelvan, and M.C. Little. Design and Implementation of a CORBA Fault-tolerant Object Group Service. In Proc. of the Second IFIP WG 6.1 International Working Conference on Distributed Applications and Interoperable Systems, DAIS99, June 1999. 24

....tion, these e#orts are limited in scope and fail to address the needs of the full diversity of group applications. This can also be said of a number of other group oriented e#orts in the CORBA world. For example Electra [17] Eternal [21] OGS [9] OFS [29] and NewTop Object Group Service [20] are all targeted at fault tolerant application scenarios and cannot easily be employed in the construction of other types of group application. Furthermore, they tend to provide flexibility though the setting of perdefined properties which, naturally enough, represent only those degrees of ....

G. Morgan, S. Shrivastava, P. Ezhilchelvan, and M. Little. Design and Implementation of a CORBA Fault-tolerant Object Group Service. In Second IFIP WG 6.1 International Working Conference on Distributed Applications and Interoperable Systems (DAIS'99), Helsinki, Finland, June 1999.

....such as virtual synchrony (VS) have been speci ed ( 1] An e cient implementation of these abstractions can be obtained if an underlying one to many TCP like multicast service 1 is available. Historically, such service has been implemented through multiple TCP streams (e.g. Newtop [8] [15] and Phoenix [9] Some other systems, like Transis ( 7] 5] are based on unreliable link layer multicast broadcast. Neither approach scales well, wasting network 1 although we use the term service , one could employ layer or abstraction as well. bandwidth and host resources, since ....

G. Morgan, S. K. Shrivastava, P. D. Ezhilchelvan and M. C. Little, Design and Implementation of a CORBA Fault-Tolerant Group Service, In 2nd IFIP WG 6.1 International Working Conference on Distributed Applications and Interoperable Services, Helsinki, June 99.

....research activities within the last years. With the request for proposal for a Fault tolerant CORBA Using Entity Redundancy [OMG98a] issued in April 1998, OMG is seeking to incorporate existing approaches for software fault tolerance into future versions of CORBA. However, most related work, i.e. [Morgan99], Felber98] Chang97] Maffeis94] implements fault tolerant behavior based on redundant execution and fault masking without taking real time constraints into account, a property absolutely needed for ensuring the safety of the devices in a tele laboratory. While the OMG has founded a Real time ....

G. Morgan, S.K. Shrivastava, P.D. Ezhilchelvan, M.C. Little; "Design and Implementation of a CORBA fault-tolerant object group service"; in Proceedings of the 2 nd IFIP WG6.1 International Working Conference on Distributed Applications and Interoperable Systems (DAIS), Helsinki, June 1999.

No context found.

G. Morgan, S. K. Shrivastava, P. D. Ezhilchelvan and M.C. Little, "Design and Implementation of a CORBA Fault-tolerant Object Group Service", Proceedings of 2 nd IFIP International Working Conference on Distributed Applications and Interoperable Systems, DAIS'99, Helsinki, June 1999.

No context found.

G. Morgan, S.K. Shrivastava, P.D. Ezhilchelvan and M.C. Little, "Design and Implementation of a CORBA Fault-tolerant Object Group Service", Distributed Applications and Interoperable Systems, Ed. Lea Kutvonen, Hartmut Konig, Martti Tienari, Kluwer Academic Publishers, 1999, ISBN 0-7923-8527-6, pp. 361-374.

No context found.

G. Morgan, S. K. Shrivastava, P. D. Ezhilchelvan and M.C. Little, "Design and Implementation of a CORBA Fault-tolerant Object Group Service", Proceedings of 2 nd IFIP International Working Conference on Distributed Applications and Interoperable Systems, DAIS'99, Helsinki, June 1999.

No context found.

G. Morgan, S.K. Shrivastava, P.D. Ezhilchelvan and M.C. Little, "Design and Implementation of a CORBA Fault-tolerant Object Group Service", Distributed Applications and Interoperable Systems, Ed. Lea Kutvonen, Hartmut Konig, Martti Tienari, Kluwer Academic Publishers, 1999, ISBN 0-7923-8527-6, pp. 361-374.

....Therefore, we propose portal replication to ensure a portal is not a single point of failure. Efforts to provide fault tolerant services via group communications [18] using Common Object Request Broker Architecture (CORBA) 22] technologies are amongst the most successful [14] 15] 16] [3]. These approaches, and in particular [16] have resulted in a fault tolerant specification for CORBA [21] CORBA s goals of interoperability through standardisation of services and protocols are similar to that of Web services. The choice, for application developers, between these two ....

....to receive, process, and reply to client requests. Group communication services have been shown to be useful in the development of applications that use replication schemes to achieve fault tolerance [14] 15] 16] We assume the availability of a group communication subsystem (NewTop [3]) that provides reliable, total ordered, delivery of multicast messages and ensures that members have a mutually consistent view of the order in which events (such as membership changes, invocations) have taken place. By reliable multicast we mean that either all or none of the functioning members ....

[Article contains additional citation context not shown here]

G. Morgan, S.K. Shrivastava, P.D. Ezhilchelvan and M.C. Little, "Design and Implementation of a CORBA Fault-tolerant Object Group Service", Distributed Applications and Interoperable Systems, Ed. Lea Kutvonen, Hartmut Konig, Martti Tienari, Kluwer Academic Publishers, 1999, ISBN 0-7923-8527-6, pp. 361-374.

No context found.

G. Morgan, S.K. Shrivastava, P.D. Ezhilchelvan and M.C. Little, "Design and Implementation of a CORBA Fault-tolerant Object Group Service", Distributed Applications and Interoperable Systems, Ed. Lea Kutvonen, Hartmut Konig, Martti Tienari, Kluwer Academic Publishers, 1999, ISBN 0-7923-8527-6, pp. 361-374.

....and [Babaoglu95a] the Relacs system. The Transis System [Dolev96] implements a partitionable membership service on a broadcast medium, and the algorithms can be found in [Amir92] Specifications in the context of multiple, overlapping groups are presented in the Newtop paper [Ezhilchelvan95] and [Morgan99] describes a CORBA compliant implementation of Newtop. Programming with Partitionable Membership Service Consider a task that needs to be done in steps S1, S2, Si, such that at least Ni processes in G must have executed a given step Si, before any process can start executing S(i 1) ....

....communication protocols can be used for implementing replica consistency protocols. Currently there is no OMG standard for an object group service. Design and development of fault tolerant group communication protocols for CORBA has therefore been a very active area of research [Felber98, Moser99, Morgan99] Three ways of incorporating object groups in CORBA have been identified [Felber98] The integration approach takes an existing group communication system and replaces the transport service of the ORB by the group service. Although this is a very efficient way of incorporating group ....

[Article contains additional citation context not shown here]

G Morgan, S K Shrivastava, P D Ezhilchelvan, and M C Little, "Design and Implementation of a CORBA Fault-Tolerant Object Group Service", Second IFIP WG 6.1 International Working Conference on Distributed Applications and Interoperable Systems (DAIS99), June 99, Helsiki, Finland.

....of a member periodically sending I am alive or NULL messages during periods it has no application level messages to send. In NewTop, after a member has neglected to send a message for a period of time, the NewTop time silence mechanism will send a I am alive message. For further details, see [13, 21]. 4.1 Related Work NewTop implements group communication services as CORBA services from scratch . In addition to being CORBA compliant, the advantage here is that the services are directly available to application builders so can be used for a variety of purposes. This approach was first ....

G. Morgan, S.K. Shrivastava, P.D. Ezhilchelvan and M.C. Little, "Design and Implementation of a CORBA Fault-tolerant Object Group Service", Distributed Applications and Interoperable Systems, Ed. Lea Kutvonen, Hartmut Konig, Martti Tienari, Kluwer Academic Publishers, 1999, ISBN 0-7923-8527-6, pp. 361-374.

....client requests nor does it send or receive episode messages. When the primary replica fails D2 assumes responsibilities of D1. 3. 5 Group Communication Object The group communication requirements of the reliability and distribution objects are satisfied by the Newtop service (a CORBA service) [24]. The Newtop service is a distributed service and achieves distribution with the aid of the Newtop Service Object (NSO) Each group member (reliability object or distribution object) is allocated an NSO. Group related communications required by a member are handled by its NSO. The Newtop service ....

G. Morgan, S.K. Shrivastava, P.D. Ezhilchelvan and M.C. Little, "Design and Implementation of a CORBA Fault-tolerant Object Group Service", Distributed Applications and Interoperable Systems, Ed. Lea Kutvonen, Hartmut Konig, Martti Tienari, Kluwer Academic Publishers, 1999, ISBN 0-7923-8527-6, pp. 361-374.

No context found.

G. Morgan et al. "Design and implementation of a CORBA fault-tolerant object group service", Proceedings of the conference on Distributed Applications and Interoperable Systems, 1999.

No context found.

Morgan, G., Shrivastava, S., Ezhilchelvan, P., Little, M.: Design and Implementation of a CORBA Fault-tolerant Object Group Service. In: Proceedings of the Second IFIP WG 6.1 International Working Conference on Distributed Applications and Interoperable Systems (DAIS'99), Helsinki, Finland (1999)

No context found.

G. Morgan, S. Shrivastava, P. Ezhilchelvan, and M. Little. Design and implementation of a CORBA fault-tolerant object group service. In Proceedings of the Second IFIP WG 6.1 International Working Conference on Distributed Applications and Interoperable Systems, Helsinki, Finland, June 1999.

No context found.

G. Morgan, S. Shrivastava, P. Ezhilchelvan, and M. Little. Design and implementation of a CORBA fault-tolerant object group service. In Proceedings of the Second IFIP WG 6.1 International Working Conference on Distributed Applications and Interoperable Systems, Helsinki, Finland, June 1999.

No context found.

Morgan G, Shrivastava S, Ezhilchelvan P, Little M. Design and implementation of a CORBA fault-tolerant object group service. Proceedings of the Second International Working Conference on Distributed Applications and Interoperable Systems, Helsinki, Finland, June 1999. Kluwer Academic Publishers, 1999; 361--374.

No context found.

G. Morgan, S.K. Shrivastava, P.D. Ezhilchelvan, and M.C. Little. Design and Implementation of a CORBA Fault-tolerant Object Group Service. In DAIS, 1999.

First 50 documents

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