Hammer, M., Shipman, D. (1980). 'Reliability Mechanisms for SDD-1: A System for Distributed Databases', ACM Transactions on Database Systems, Vol. 5, no. 4, (December), pp. 431-466.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....was proposed that maintains the consistency in spite of network partitions. In [9] enhanced 3PC protocol is proposed which is more resilient to network partitioning failures than quorum based 3PC protocol. In order to deal with the failure of the coordinator, backup processes are used in SDD 1[7]. These processes are initiated by the coordinator before initiating the commit protocol and substitute the coordinator in case of its failure. In order to ensure that only one process will substitute for the coordinator, backups are linearly ordered, so that the first one looks at the ....

....second phase, the coordinator sends the updates to participants. In third phase, the coordinator communicates its decision to backups. And in fourth phase the coordinator sends its decision to participants. In this paper we use the notion of backup site similar to the notion of backup process in [7]. However, in BC protocol, in case of coordinator s failure the backup site does not assume the role of coordinator. Also, there is no periodic exchange of control messages between the coordinator and corresponding backup site. Instead, the sites themselves resolve the blocking by contacting the ....

Michael Hammer and David Shipman, "Reliability mechanisms for SDD-1: A system for distributed databases", ACM Transactions on Database Systems, vol.5, no.4, December 1980, pp.431-466.

....and the set of copies in current use is indicated by a replicated set of directories analogous to the configuration state. The available copies algorithm differs from ours in an impOrtant respect: as noted by its authors, it fails to preserve consistency in the presence of partitions. SDD.1 [Hammer 80] employs a similar replication scheme in which sites that have crashed are configured out of the system. It, too, does not tolerate communication failures or partitions. These examples are 128 discus.seal further in the next chapter, where we propose a reelhod Ior enhancing availability in ....

Hammer, M. M., and Shipman D. W. "Reliability Mechanisms in SDD-1, a System for Distributed Databases", ACM Transactions on Database Systems 5,4, December 1980, 431.486.

.... undo) algorithm is also based on methods used in many systems (e.g. Gray78, IMS78, SMB79, Paxton79, LeLann81] The distributed transaction management algorithm is original, except in its incorporation of a two phase commit protocol [Gray78, Gray80a, Paxton79, IMS78, Reed78, LeLann81, Lindsay79, HS80] The distributed deadlock detection algorithm is an extension of work reported in [Goldman77] and [Obermarck80] Some of the methods of presentation that we have used derive from other specific works. In particular, the system model and failure model of Chapter 2 was inspired by [LS] and the ....

....will decide whether to complete or abort. There are two underlying assumptions here: that no nodes crash forever, and that eventually any given pair of nodes can communicate. The latter assumption is fairly reasonable; if necessary, certain messages could be spooled through third parties (see [HS80] for example) However, it is not quite so reasonable to assume that nodes may not fail permanently. We will return to this point in a moment. There are some other facts to note about the algorithm. First, though it may seem slightly strange for a participant to respond completed when it ....

[Article contains additional citation context not shown here]

Michael Hammer and David Shipman, "Reliability Mechanisms for SDD-1: A System for Distributed Databases", ACM Trans. on Database Sys., Vol. 5, No. 4, Dec. '1980, pp. 431-466.

....caused by network partitions, node crashes, or byzantine agent errors) Consequently, if agents can generate success or crash events, no timeout mechanism is needed. However, crash event management is very hard to accomplish. The necessary surveillance protocols are very complex (see [Walter82] [HamShi80]) and do not consider migrating elements. Furthermore, mobile devices are hard to surveil due to their sporadic connection to the rest of the network. In contrast to OR groups, AND groups succeed only if all agents have accomplished their task. AND groups are well suited for various scenarios ....

M. Hammer, D. Shipman. "Reliability Mechanisms for SDD-1: A System for Distributed Databases", in: ACM Transactions on Database Systems 5:4, December 1980.

No context found.

Hammer, M., Shipman, D. (1980). 'Reliability Mechanisms for SDD-1: A System for Distributed Databases', ACM Transactions on Database Systems, Vol. 5, no. 4, (December), pp. 431-466.

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