L. E. Moser, P. M. Melliar-Smith and P. Narasimhan, "A Fault Tolerance Framework for CORBA," Proceedings of the IEEE International Symposium on Fault-Tolerant Computing, Madison, WI (June 1999), pp. 150-157.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....Strategy the client request parameters to alter the behavior of the application or to enhance the application with new functionality. The modified requests are then mapped onto a reliable group communication messaging system. The interception strategy is used heavily in the Eternal system [10, 48] to construct a reliable ORB environment. Eternal intercepts system calls made by the client object through the lower level I O subsystem and maps these system calls to the reliable multicast subsystem. Eternal differs from Orbix Isis or Electra in that Eternal does not modify the ORB or the ....

Louise Moser, P. Melliar-Smith, and Priya Narasimhan, "A Fault Tolerance Framework for CORBA," in International Symposium on Fault Tolerant Computing, Madison, WI, June 1999, pp. 150--157.

....is monitored and categorized, which marks the difference from our work. Our approach tries to manifest robustness failures in the CORBA ORB native code using exceptional inputs. Various efforts have been made to build fault tolerant CORBA applications, CORBA services and middleware, such as [12][13][20] The Fault Tolerant CORBAStandard [16] extends CORBA for applications requiring high dependability, attempting to eliminate sources of single point failures. This standard mainly aims at tolerating crash failures using replication and does not address issues of exceptional parameter input ....

L. Moser, P. Melliar-Smith and P. Narasimhan, "A Fault Tolerance Framework for CORBA". Proceedings of FTCS 29, Madison, Wisconsin, June 15-18, 1999.

....appropriately onto properties of the replicated resources. This mapping is often not straightforward, especially when some of the QoS requirements may be conflicting. For example, in order to provide good fault tolerance, we could allocate all the available replicas to service a client (e.g. [4, 25, 47, 11, 55, 15]) However, such an approach would not be scalable, as it would increase the load on all the replicas and result in higher response times for the remaining clients. On the other hand, assigning a single replica to service each client would allow multiple clients to be serviced concurrently. ....

....the accesses result in modifications to the replicated state. In order for the responses to be meaningful to the clients, it is important to bound the degree of inconsistency when the replicated information is time varying. Traditionally, replicated systems provide either strong consistency (e.g. [55, 63, 67]) or weak consistency (e.g. 77, 37, 21] guarantees for replicated data. Strong replica consistency requires that at the end of each method execution by a replicated object, all the replicas of the object have the same state. Weak consistency specifies that the replicated state will eventually ....

[Article contains additional citation context not shown here]

L. Moser, P. Melliar-Smith, and P. Narasimhan. A Fault Tolerance Framework for CORBA. In Proc. of the IEEE Intl. Symp. on Fault-Tolerant Computing, pages 150--157, June 1999.

.... an ORB to retarget its requests) or an agent (to restart a task, or reboot a node altogether) The agents and managers are described in this paper as if they are free standing processes, even though in practice some of these components may be combined with other components in a dependability ORB [18, 19], or an application management system [20] Failures of system entities are detected by mechanisms such as heartbeats, timeouts on periodic polls, and timeouts on requests between application tasks. Heartbeat messages from an application task can be generated by a special heartbeat interrupt ....

L. E. Moser, P. M. Melliar-Smith and P. Narasimhan, "A fault tolerance framework for CORBA", Proc. of 29th Annual Int. Symposium on Fault-Tolerant Computing, 1998, pp. 150-157.

....reliable distributed computing. For example, the AQuA project [Ren01, CRS 98] at UIUC uses the Ensemble group communication system to build a middleware that can provide applications with the desired level of dependability in the presence of crash failures and value faults. The Eternal System [MMSN99] at UCSB uses the Totem group communication system to extend CORBA with object replication and fault tolerance. We now describe some group communication systems that are related to our research. IIOP is the Internet Inter Orb Protocol, which specifies transfer syntax and message format to ....

Louise Moser, P. Michael Melliar-Smith, and Priya Narasimhan. A fault tolerance framework for CORBA. In Proceedings of the IEEE International Symposium on Fault-Tolerant Computing, pages 150--157, Madison, WI, June 1999.

....that can together meet the client s timing constraints with at least the probability requested by the client. We conclude with experimental results based on our implementation. 1. Introduction Server replication is a popular approach for building fault tolerant distributed services (e.g. [1, 7, 12, 3, 14, 6]) Replication is also a commonly used solution for improving the scalability of a distributed service, i.e. to ensure that the response time of a service does not significantly degrade with an increase in the number of clients accessing the service (e.g. 10, 13, 4] Achieving both fault ....

L. E. Moser, P. M. Melliar-Smith, and P. Narasimhan. A Fault Tolerance Framework for CORBA. In Proceedings of the IEEE International Symposium on Fault-Tolerant Computing, pages 150--157, June 1999.

....contract F30602 98 C0187. consistency models have been studied extensively. In the strong consistency model, concurrent operations on replicated data are equivalent to a sequential execution on nonreplicated data. Pessimistic replication algorithms, such as active and passive replication (e.g. [1, 11, 8, 12]) have traditionally been used to maintain strong consistency among replicated data. Although these algorithms, which provide single copy semantics, ensure correctness for a wide class of applications (e.g. banking transactions) the performance overheads incurred in maintaining mutually ....

L. Moser, P. Melliar-Smith, and P. Narasimhan. A Fault Tolerance Framework for CORBA. In Proc. of the IEEE Intl. Symp. on Fault-Tolerant Computing, pages 150--157, June 1999.

....group 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 ....

....group communication system; here (IIOP) messages issued by an ORB are intercepted and mapped on to calls of the group communication system. The best known example of this approach is the Eternal system for object replication that makes use of Totem group communication system in this manner [Moser99] The major advantage of this approach is that no modifications to the ORB are required. The shortcomings are that the approach is only possible if the host operating system permits interception (Unix in the case of Eternal) secondly, as the group communication system is not available to CORBA ....

L. E. Moser, P.M. Melliar-Smith and P. Narasimhan, "A fault tolerance framework for CORBA", Proc. of Fault tolerant Computing Symp., FTCS-29, June 1999, pp. 150-157.

....for example, clients that access a crashed or failing server can be migrated to other servers until the failure is resolved. Load balancing services need not provide full fault tolerance capabilities, however, i.e. it should not be the role of a load balancing service to detect and mask failures [19, 20]. Instead, they should provide mechanisms to handle those failures efficiently when they are detected by administrators or other components in the system. Support administrative tasks: System administrators may need to add new object replicas dynamically, without disrupting or suspending service ....

L. Moser, P. Melliar-Smith, and P. Narasimhan, "A Fault Tolerance Framework for CORBA," in International Symposium on Fault Tolerant Computing, (Madison, WI), pp. 150--157, June 1999.

....specific recovery policy. For instance, some object groups have less demanding dependability requirements and may tolerate a weaker recovery policy than other groups. The properties of our replication management framework described above, enables autonomous replication management. The Eternal [6] and DOORS [7] implementations of the Fault Tolerant CORBA specification [9] provide a similar recovery facility, however, they do not allow application specific recovery policies nor do they autonomously distribute replicas. Jgroup is a GCS for partitionable environments that integrates the ....

L. E. Moser, P. M. Melliar-Smith, and P. Narasimhan. A Fault Tolerance Framework for CORBA. In Proceedings of the IEEE International Symposium on FaultTolerant Computing, pages 150--157, Madison, WI, June 1999.

....is monitored and categorized, which marks the difference from our work. Our approach tries to manifest robustness failures in the CORBAORB native code using exceptional inputs. Various efforts have been made to build fault tolerant CORBA applications, CORBA services and middleware, such as [12][13][20] The Fault Tolerant CORBAStandard [16] extends the present CORBAstandard for applications requiring high dependability, targeting zero single point failures. This standard mainly aims at tolerating crash failures using replication and does not address issues of exceptional parameter input ....

L. Moser, P. Melliar-Smith and P. Narasimhan, "A Fault Tolerance Framework for CORBA". Proceedings of FTCS 29, Madison, Wisconsin, June 15-18, 1999.

....by some polling mechanism, this exception can be raised even before the POA tries to make an upcall. The permanent flag can be 4 used to set the replicated IOR as the one to which all future invocations will be made, thereby providing the building blocks to improve application fault tolerance [9, 10]. Multiple interceptors: Multiple interceptors can be registered with the ORB for each interception point. This type of interceptor is handled according to the following rules: # Only one starting interception point can be called for an invocation; # Only one ending interception point can be ....

L. Moser, P. Melliar-Smith, and P. Narasimhan, "A Fault Tolerance Framework for CORBA," in International Symposium on Fault Tolerant Computing, (Madison, WI), pp. 150--157, June 1999.

No context found.

L. E. Moser, P. M. Melliar-Smith and P. Narasimhan, "A Fault Tolerance Framework for CORBA," Proceedings of the IEEE International Symposium on Fault-Tolerant Computing, Madison, WI (June 1999), pp. 150-157.

No context found.

L. E. Moser, P. M. Melliar-Smith and P. Narasimhan, "A Fault Tolerance Framework for CORBA," Proceedings of the IEEE International Symposium on Fault-Tolerant Computing, Madison, WI (June 1999), pp. 150-157.

No context found.

L. E. Moser, P. M. Melliar-Smith and P. Narasimhan, "A Fault Tolerance Framework for CORBA," Proceedings of the IEEE International Symposium on Fault-Tolerant Computing, Madison, WI (June 1999), pp. 150-157.

No context found.

L. Moser, P. Melliar-Smith, and P. Narasimhan. A Fault Tolerance Framework for CORBA. In Proc. of the IEEE Intl. Symp. on Fault-Tolerant Computing, pages 150--157, June 1999.

No context found.

Louise Moser, P. Michael Melliar-Smith, and Priya Narasimhan. A fault tolerance framework for CORBA. In Proceedings of the IEEE International Symposium on Fault-Tolerant Computing, pages 150--157, Madison, WI, June 1999.

No context found.

L.E. Moser, P.M. Melliar-Smith and P. Narasimhan. "A fault tolerance framework for CORBA," in Proc. of the 29th Fault Tolerant Computing Symposium, pp. 150-157, Madison, USA, June 1999.

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