P. Krishna, N.H. Vaidya, and D.K. Pradhan. "Recovery in Distributed Mobile Environments". IEEE Workshop on Advances in Parallel and Distributed System, Oct. 1993.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....wireless link usually suffers with a high bit error rate, a little bandwidth, and a long transfer delay. Mobile terminals even disconnect from the hosting network intermittently [23] Mobile systems are more often subject to environmental conditions which can cause loss of communications or data [10]. All of these cause transient failures more frequent. Thus, mobile computing requires techniques to provide fault tolerance to continue its services despite such permanent and transient failures. OMG has specified Fault Tolerant CORBA (FT CORBA) 18] as a standard. FT CORBA is based on entity ....

P. Krishna, N. H. Vaidya, and D. K. Pradhan. Recovery in distributed mobile environments. Proceedings of the IEEE Workshop on Advances in Parallel and Distributed Systems, pages 83--88, October 1993.

....if so, how efficient do they perform This problem has attracted a lot of researchers interest. Issues that have been considered on mobile distributed computing environment include causal ordering of messages [2, 8] fault tolerance [1] distributed snapshot [12] and distributed checkpointing [5, 9]. The rest of this paper is organized as follows. The formal definition of distributed termination detection, as well as its applications, is presented in Section 2. In Section 3, we review two existing termination detection protocols and discuss the possible problems to use them directly on a ....

P. Krishna, N. Vaidya, and D. Pradhan. Recovery in distributed mobile environments. In Proc. Workshop Advances in Parallel and Distributed Systems, pages 83--88, 1993.

....the undelivered updates of the transaction to the mailbox. The new agent then logs itself into the mailbox as the current BSA, so all the new updates to the transaction are routed to the correct BSA. If the MH crosses zones during the transmission of a transaction, then the Pessimistic hand off [14] strategy is used. The Pessimistic hand off strategy is designed for applications where long service disruptions are not tolerated. When a MH crosses from zone A to zone B, the hand off occurs. The BSA of zone A sends the portion of the transaction that it stored in its temporary storage ....

P. Krishna, N.H. Vaidya, and D.K. Pradhan. Recovery in Distributed Mobile Environments. In IEEE Workshop on Advances in Parallel and Distributed Systems, pages 83--88, 1993.

....design of distributed computation. This problem has attracted a lot of researchers interest. Issues that have been considered on mobile distributed computing environment include causal ordering of messages [2, 15, 22] fault tolerance [1] distributed snapshot [21] and distributed checkpointing [9, 16]. In this paper, we study one fundamental problem in distributed computing, the termination detection problem [17] A distributed system can be generally defined as a set of autonomous processes which cooperate with each other to complete a job. To coordinate computation and exchange data, ....

P. Krishna, N. Vaidya, and D. Pradhan. Recovery in distributed mobile environments. In Proc. Workshop Advances in Parallel and Distributed Systems, pages 83--88, 1993.

....hand held mobile hosts, such laptop PCs, This research is supported by the National Science Council of the Republic of China under Grant # NSC882213 E 008 014 palmtop PCs, and PDAs are also available in the market. Mobile computing (or nomadic computing) has received intensive attention recently [1, 2, 7, 9, 13, 14, 15, 16]. There are two possibilities to form a wireless network: infrastructure and ad hoc. In infrastructure networks, a number of base stations are used, through which all communications to or from the mobile hosts must go. The based stations can then be interconnected by wired networks, but generally ....

P. Krishna, N. Vaidya, and D. Pradhan. Recovery in Distributed Mobile Environments. In Proc. Workshop Advances in Parallel and Distributed Systems, pages 83--88, 1993.

....cost. One way to achieve these is by storing a part (or all) of the state information related to a mobile host in its mobile support station. For example, the algorithm of Acharya and Badrinath [2] for multicasting a message to several mobile hosts and the algorithms of Krishna et. al [9] (to recover from a mobile host failure) store state information related to a y This research was supported in part by NSF under Grant No. CCR 9110177, by the Texas Advanced Technology Program under Grant No. 9741 036, and by grants from Alcatel Network Systems. 1 mobile host in its support ....

....h occasionally. To recover from failure, h first gets its recent checkpointed state, and replays the messages stored after the recent checkpointed state. Mobile host h processes the messages to recover to its state prior to failure. This approach is similar to the recovery scheme of Krishna et. al [9] for mobile host failure. Another approach is to have the complete state information of mobile host h at pri mss(h) When a message m is sent to mobile host h, the state of h changes. pri mss(h) also updates state of h (stored locally) using the message m. Thus the state of h stored at pri mss(h) ....

Krishna, P., Vaidya, N., and Pradhan, D. Recovery in distributed mobile environments. In Proceedings of the IEEE Workshop on Advances in parallel and distributed systems (1993), pp. 83--88.

....so, how efficient do they perform This problem has attracted a lot of researchers interest. Issues that have been considered on mobile distributed computing environment include causal ordering of messages [2, 19, 26] fault tolerance [1] distributed snapshot [25] and distributed checkpointing [13, 20]. In this paper, we study one fundamental problem in distributed computing, the termination detection problem [21] A distributed system can be generally defined as a set of autonomous processes which cooperate with each other to complete a job. To coordinate computation and exchange data, ....

P. Krishna, N. Vaidya, and D. Pradhan. Recovery in distributed mobile environments. In Proc. Workshop Advances in Parallel and Distributed Systems, pages 83--88, 1993.

....its pushed items are followed by a marker. As transactions are processed, their effects are logged on the mobile host where the transaction executes to facilitate recovery from failed or aborted transactions. Mobile host logs are checkpointed periodically on an appropriate mobility support station [18, 3]. Merging of Fragments When a mobile host reconnects or the stack partition is no longer needed at the mobile host, any stack fragments remaining in the stack partition must be reconciled with master stack on the database server. The stack partition along with the log for committed transactions ....

Krishnan P., N. Vaidya and D. Pradham. Recovery in Distributed Mobile Environments. In Proc. of IEEE Workshop on Advances in Parallel and Distributed Systems, pp. 83-88, 1993.

.... to the change in physical connectivity, resource constraints of mobile hosts and limited bandwidth of the wireless links [1] This has spawned considerable amount of research in mobile computing: designing communication protocols [3, 9] file system operations [4] and providing fault tolerance [5]. In this paper, we consider the problem of providing a particular kind of communication support, namely, causally ordered message delivery to mobile hosts. Consider two messages m and m 0 sent to the same destination such that sending of m happened before sending of m 0 . Causal ordering ....

Krishna, P., Vaidya, N., and Pradhan, D. Recovery in distributed mobile environments. In Proceedings of the IEEE Workshop on Advances in Parallel and Distributed Systems (1993), pp. 83--88.

....hosts without such resources simply cease to function if disconnected. There are a variety of approaches for dealing with disconnection Many ideas can be borrowed from fault tolerance, such as replication of the mobile host state and atomic multicast to ensure that multiple hosts receive messages [VKP93, PKV96] However, since failures are more common, any technique must be inexpensive, as well as not interfere with the rest of the system. Other solutions are specific to mobile computing, such as cases where routing protocols redirect messages destined for a disconnected host to a proxy that ....

N. H. Vaidya, P. Krishna, and D. K. Pradhan. Recovery in distributed mobile environments. In Proceedings of IEEE Workshop on Advances in Parallel and Distributed Systems, pages 83--88. IEEE, October 1993.

....moves between cellular zones. The mobile support router hand off and zone crossing problem are addressed using a mobile protocol adapted from the Columbia mobile host protocol proposal [11, 12, 10, 17] 3.2 MSR Hand off An MH makes the decision when an MSR hand off should occur. The hand off [14] criteria used is typically a physical layer issue and will not be discussed in this paper. When the MH decides to switch between MSRs, the Greeting packet sent to the target MSR must contain a list of MSR IP addresses which previously serviced the MH during its current connection within the ....

P. Krishna, N. Vaidya, and D. Pradhan. Recovery in Distributed Mobile Environments. In IEEE Workshop on Advances in Parallel and Distributed Systems, pages 83--88, 1993.

....extensions to existing traditional recovery schemes to suit the mobile environment. We divide recovery schemes into state saving and handoffs. We have proposed strategies for state saving and handoffs. The recovery schemes use one combination of state saving and handoff throughout the execution [13]. 4.3.1 Performance Metrics Trade off in recovery schemes is between the failure free overhead and the recovery overhead. The performance of a recovery scheme in a mobile wireless network depends mainly on the (i) cost of wireless transmission, ii) communication mobility ratio of the user, and ....

....routing overhead. We are currently analyzing the overhead due to cluster maintenance and topology updates. We are also currently investigating schemes to tolerate corrupt wireless links. 5. 3 Fault Tolerance in Mobile Wireless Networks We have developed strategies for state saving and handoffs [13]. We are currently analyzing the proposed strategies. Analysis and simulations will be used to determine the environments where a particular recovery scheme is best suited. ....

P. Krishna, Nitin H. Vaidya and D. K. Pradhan, "Recovery in distributed mobile environments," IEEE Workshop on Advances in Parallel and Distributed Systems, pp. 83-88, Oct. 1993.

....as handoff, is transparent to the mobile host. Thus, end to end connectivity in the dynamically changing network topology is preserved transparently. A mobile host may become unavailable due to (i) failure of the mobile host, ii) disconnection of the mobile host, and (iii) wireless link failure [7, 8, 11, 12, 16]. Loss of battery power make disconnections from the network frequent, and sometimes unpredictable. Because of their frequency, disconnections must be treated differently than failures. The difference between disconnection and failure is its elective nature. User initiated disconnections can be ....

....cost State saving cost, Perf. metrics Recovery cost Recovery cost, Handoff time Table 1: Differences between Static Wired and Mobile Wireless Networks: Recovery Perspective It will now be discussed why traditional fault tolerance schemes cannot be applied to a mobile wireless environment [11, 12, 16]. Some of the differences between static and mobile networks are enumerated in Table 1. Traditional fault tolerance schemes like checkpointing and message logging [9, 15] require a stable storage for saving the checkpoint and the logs. It has been pointed out [3] that while the disk storage on a ....

P. Krishna, Nitin H. Vaidya and D. K. Pradhan, "Recovery in distributed mobile environments, " IEEE Workshop on Advances in Parallel and Distributed Systems, pp. 83-88, Oct. 1993.

....harsh environment that a portable computer faces [3] This will require any critical data (e.g. database logs, process checkpoint) to be stored elsewhere other than the mobile computer. This in turn will impact performance of protocols (e.g. recovery protocols) that require these critical data [1, 48, 65]. ffl Disconnections : Mobile units run on batteries; with limited capacity [20, 31] Limitations in battery power and bandwidth make disconnections from the network very frequent. Disconnections have various degrees depending on bandwidth availability. Because of their frequency, disconnections ....

....mobile host failure in an infrastructure mobile wireless network. The system model of the infrastructure network is the same as explained in Chapter I. A mobile host may become unavailable due to (i) failure of the mobile host, ii) disconnection of the mobile host, and (iii) wireless link failure [20, 31, 48, 65]. Loss of battery power make disconnections from the network frequent, and sometimes unpredictable. Because of their frequency, disconnections must be treated differently than failures. The difference between disconnection and failure is its elective nature. User initiated disconnections can be ....

[Article contains additional citation context not shown here]

P. Krishna, Nitin H. Vaidya, D. K. Pradhan, "Recovery in Distributed Mobile Environments," Proc. IEEE Workshop on Advances in Parallel and Distributed Systems, pp. 83-88, October, 1993.

....parameters apart from the failure rate of the mobile host also play a key role in determining the effectiveness of a recovery scheme. We identify the tradeoff parameters for a recovery scheme in a mobile environment. We analyze the performance of the recovery schemes proposed in this report and [7]. We also determine those mobile environments where a particular recovery scheme is best suited. Other variations of the recovery schemes presented here are also plausible. Our main aim is to present the limitations of the new mobile computing environment, and its effects on recovery protocols. In ....

.... 3. For a given environment, which recovery scheme is preferable This report is organized as follows. Section 2 presents a review of related work. Section 3 presents the recovery strategies. For sake of better understanding we also give a brief overview of the strategies presented in [7]. Section 4 presents the performance analysis of the recovery strategies, and conclusions are presented in Section 5. 2 Related Work In the recent years, there has been a lot of work [4] dealing with mobility management, database system issues, network protocols, disconnected operation and ....

[Article contains additional citation context not shown here]

P. Krishna, Nitin H. Vaidya and D. K. Pradhan, "Recovery in Distributed Mobile Environments, " IEEE Wkshp on Advances in Parallel and Distributed Sys., pp. 83-88, Oct. 1993.

....a search. We build analytical models to compare the performance of the proposed approach with the IS 41 scheme [5] Apart from location management, we are also working on developing routing protocols in ad hoc wireless local area networks [6] and recovery protocols for mobile wireless networks [7]. Acknowledgments We thank the reviewers for their valuable comments. ....

P. Krishna, Nitin H. Vaidya and D. K. Pradhan, "Recovery in distributed mobile environments," IEEE Workshop on Advances in Parallel and Distributed Systems, pp. 83-88, Oct. 1993.

No context found.

P. Krishna, N.H. Vaidya, and D.K. Pradhan. "Recovery in Distributed Mobile Environments". IEEE Workshop on Advances in Parallel and Distributed System, Oct. 1993.

No context found.

P. Krishna, N.H. Vaidya, and D.K. Pradhan. "Recovery in Distributed Mobile Environments". IEEE Workshop on Advances in Parallel and Distributed System, Oct. 1993.

No context found.

Krishna, P., Vaidya, N. H., and Pradhan, D. K. Recovery in distributed mobile environments. In IEEE Workshop on Advances in Parallel and Distributed Systems (October 1993).

No context found.

P. Krishna, N. H. Vaidya, and D. K. Pradhan. Recovery in distributed mobile environments. Proceedings of the IEEE Workshop on Advances in Parallel and Distributed Systems, pages 83--88, October 1993.

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