S. B. Davidson, H. Garcia-Molina, and D. Skeen, "Consistency in Partitioned Networks," ACM Comp. Surveys, vol. 17, no. 3, Sept., 1985.  Home/Search   Document Not in Database   Summary   Related Articles   Check

.... procedure call [5] the use of timeouts, and the use of endto end arguments in placement of functionality [6] Fault tolerance, including atomic transactions, distributed and nested transactions, and two phase commit [7] High availability, including optimistic and pessimistic replica control [8], mirrored execution [9] and optimistic recovery [10] Remote information access, including caching, function shipping, distributed file systems, and distributed databases [11] Security, including encryption based mutual authentication and privacy [12] Mobile Computing The appearance of ....

S. B. Davidson, H. Garcia-Molina, and D. Skeen, "Consistency in Partitioned Networks," ACM Comp. Surveys, vol. 17, no. 3, Sept., 1985.

....A distributed database may be replicated to enhance the availability of the data. Then, ensuring correctness and consistency of the data is important. Increasing the availability of a service is usually somewhat easier, but the integrity of the service must be retained. Davidson states in [17] that availability and correctness are competing goals of a faulttolerant distributed system. To achieve higher availability of the supplied services in the case of network partitioning, all nodes can be allowed to perform transactions as usual. These transactions may produce incorrect results and ....

Davidson, Susan B. et al.: "Consistency in partitioned networks", ACM Computing Surveys, Vol. 17 (3), 1985, pp. 341-370.

....e.g. the Eternity Service proposed in [9] 5.3. 5 Consistency minority partition, whether it is safe to proceed with a task that alters shared data [51] Traditional, pessimistic consistency control algorithms strive to offer single copy semantics, i.e. users see only a single copy at all times [84, 34]. Such strong consistency requirements are therefore impossible or very expensive to meet in our target environment, where full connectivity might never happen. The consequence of this is that merging of conflicting updates may be required upon reconnection. Whether concurrent updates represent ....

DAVIDSON, S. B., GARCIA - M OLINA, H., AND SKEEN, D. Consistency in partitioned networks. ACM Computing Surveys 17, 3 (Sept. 1985), 341--370. 149

....of participants, and the second phase propagates information to a majority. A quorum system [13] a generalization of majority sets is a collection of sets such that any two sets, called quorums, intersect [12] Quorum systems have been used to implement data replication protocols, e.g. [4,6,14]. Consensus algorithms have been used as building blocks in other work, e.g, 15] Note. This paper is an extended abstract of a full report [20] The full version includes specifications of all components, complete proofs, and additional results. 2 Data Types We assume distinguished ....

S.B. Davidson, H. Garcia-Molina and D. Skeen, "Consistency in Partitioned Networks", ACM Computing Surveys, vol. 15, no. 3, pp. 341-370, 1985.

....can obviously lead to inconsistencies in the various cached copies of information. A pessimistic approach that only allows one copy to be updated will avoid inconsistencies; an optimistic approach allows more than one copy to be updated but then conflicts have to be resolved if they occur [3]. The Coda file system uses an optimistic approach and experience has shown that inconsistencies rarely occur [18] For simplicity, and to focus on the security aspects in databases, we will use a pessimistic approach in this paper. Obviously this limits availability of an object to one mobile ....

S B Davidson, H Garcia-Molina and D Skeen, "Consistency in Partitioned Networks", ACM Computing Surveys, 17(3):341--370, 1985

....of the main issues involved in the management of data in largescale and mobile computing environments is the handling of uncoordinated independent concurrent modifications. In database systems, where it has been most actively studied, concurrency control is usually achieved through transactions [4]. Transactions may be implemented either using pessimistic (e.g. locks) or optimistic (e.g. timestamps) techniques. While using optimistic techniques, several transactions are allowed to proceed concurrently until commit time, being aborted if ACID properties have been violated. As ACID properties ....

....objects, with concurrent updates, either discarding conflicting updates, creating multiple version, merging conflicting updates [21] or executing updates transformations [7] RELATED WORK Several systems have been developed to manage data in large scale environments. Notably, database systems [4], based on transactions, define a widely used and wellunderstood model of concurrency control. Some systems [20] have even introduced extensions to support disconnected operation. However, as we have already discussed, transactional techniques are not suitable for asynchronous groupware. Lotus ....

S. Davidson, H. Garcia-Molina, D. Skeen. Consistency in Partitioned Networks. ACM Computing Surveys, C31, 1982.

....from concurrent to sequential t ab TR 2002 04, University of Stuttgart, IPVS, Germany 21 Strong consistency based on the concept of serializability [HR83] has been addressed in the domain of distributed databases extensively. Since, however, consistency is a trade off to availability [DGS85], strong consistency may result in poor availability if the presence of frequent network partitioning. Therefore, weaker consistency levels have been proposed to increase the availability of data. The seminal work in [Dem 87] proposes epidemic algorithms to update copies in fixed networks. Their ....

....the most current state of a data item and accepting a weaker consistency of replicas in order to increase availability under real time conditions. However, it differs form our in the assumption that there exists a single update source for each object. The replication protocols in [KC00] [DGS85], Dem 87] provide means for synchronizing the copies in rejoined partitions. While our algorithms ensure that updates are applied according to our consistency constraints, we rely on the underlying (hyper) flooding mechanism to rebroadcasts states that are not available in each of the rejoined ....

Davidson, S. B., Garcia-Molina, H., Skeen, D.: "Consistency in Partitioned Networks", ACM Computing Surveys, Vol. 17, No. 3, September 1995

....sets such that any two quorums intersect [20] Another approach is to classify quorums as read quorums and write quorums such that any read quorum intersects any write quorum, and (sometimes) such that any two write quorums intersect. Quorums have been used to implement data replication protocols [2, 8, 9, 11, 16, 17, 22, 23]. Consensus algorithms have been used as building blocks in other work, e.g, 27] Paper organization. The rest of the paper is organized as follows. Section 2 describes some data types used by our algorithms. Section 3 contains our speci cation for the Rambo recon gurable atomic memory ....

S.B. Davidson, H. Garcia-Molina and D. Skeen, \Consistency in Partitioned Networks", ACM Computing Surveys, vol. 15, no. 3, pp. 341-370, 1985.

....two sets S; R 2 S have a nonempty intersection. Set systems with the intersection property are known as quorum systems, and the sets in such a system are called quorums. Quorum systems have been used in the study of problems such as mutual exclusion (cf. 39] data replication protocols (cf. [7, 18]) name servers (cf. 32] selective dissemination of information (cf. 46] and distributed access control and signatures (cf. 34] A protocol template based on quorum systems works as follows. In order to perform some action (e.g. update the database, enter a critical section) the user ....

S. B. Davidson, H. Garcia-Molina, and D. Skeen, Consistency in partitioned networks, ACM Computing Surveys, 17 (1985), pp. 341--370.

....are based on the premise that if a problem or inconsistent state can be detected before it causes permanent harm, then all work to correct a problem or maintain consistent state can be delayed until it is known to be needed. They operate using the same lazy philosophy as optimistic algorithms [23, 24]. The Dynamic Source Routing protocol (DSR) completely avoids periodic behavior, and uses source routing to solve the routing information consistency problem. First, DSR is completely on demand, which allows the overhead of the protocol to automatically scale directly with the need for reaction ....

Susan B. Davidson, Hector Garcia-Molina, and Dale Skeen. Consistency in Partitioned Networks. ACM Computing Surveys, 17(3):341--370, September 1985.

....important design issue in replication is how replicas are presented to users. Traditional pessimistic replication algorithms offer single copy semantics, that is, they give users an illusion of having only a single, highly available copy of an object by keeping the replicas identical all the time [6, 17]. They are called pessimistic algorithms, because they prohibit accesses to a replica unless the replica contents are provably up to date. Although these algorithms are essential in a class of applications, such as banking, that must give correct answers all the time at all cost, they have one ....

....is to combine two algorithms, an efficient algorithm for the common case and a slower but fault tolerant algorithm for the emergency case. Examples include the use of unreliable multicast protocols [18, 47] Section 3.2. 1) and the combination of distributed transactions and optimistic replication [17]. This line of algorithms, however, has failed to gain popularity due to its complexity. The second approach, which has yet to see an investigation, is optimizing existing optimistic algorithms. For example, algorithms based on timestamp vectors use an update log that must support a complicated ....

Susan B. Davidson, Hector Garcia-Molina, and Dale Skeen. Consistency in partitioned networks. ACM Computing Surveys, 17(3):341--370, 1985.

.... call [3] the use of timeouts, and the use of end to end arguments in placement of functionality [28] fault tolerance, including atomic transactions, distributed and nested transactions, and two phase commit [13] high availability, including optimistic and pessimistic replica control [9], mirrored execution [4] and optimistic recovery [37] remote information access, including caching, function shipping, distributed file systems, and distributed databases [30] security, including encryption based mutual authentication and privacy [23] 2 2.2. Mobile Computing The ....

Davidson, S.B., Garcia-Molina, H., Skeen, D. Consistency in Partitioned Networks. ACM Computing Surveys 17(3), September, 1985.

....or via caching servers that are either globally known or can be discovered. Localized caching is of no interest here. Tight consistency constraints can be thought of as strict serializability in the replicated database sense. In many environments, much looser constraints may be adequate [14]. For example, it may be adequate to ensure that a user only sees one of the last k versions perhaps with the added constraint that the version seen by a user can never go backwards. In a peer to peer environment, the protocol savings due to weaker consistency may be well worth the precision ....

S.B. Davidson, H. Garcia-Molina and D. Skeen, "Consistency in Partitioned Networks", ACM computing surveys, Vol 17, No 3, Sept 1985.

.... tends to open a Pandora s box of complexity; the many algorithms that have been used for propagating updates and reconciling inconsistent changes made to different replicas are complicated, and it is difficult to give a precise definition or make guarantees on what sort of consistency they provide [9]. We have cut through this knot of complexity by restricting the mutability of the data to be replicated, allowing us to give a simple, flexible, and comprehensible definition of consistency for partial replicas and to preserve this consistency with simple algorithms. Our approach to replication ....

Susan B. Davidson, Hector Garcia-Molina, and Dale Skeen. Consistency in partitioned networks. ACM Computing Surveys, 17(3):341--370, September 1985.

....take the view that this code is not to be trusted, and take measures to protect the system and its data from malicious or erroneous application code. Bayou merge procedures, which are roughly analogous to reconcile methods, are not allowed to have any side effects other than writing the database [Dem94]. This restriction protects a device hosting against arbitrary actions by a merge procedure, which is, in effect, a mobile agent 36 of the application writer. A fundamental tenet of Coda is that server machines shared assets requiring strong protection are trusted, and client ....

....state, but by the act of synchronization. In particular, it is possible for a sync store to be synchronized with replicas other than those in its registry. There is no mechanism by which a sync store announces that it has joined the network, or that it is leaving. Bayou supports fluid replication [Dem94], in which individual users can create new replicas without any central registration. However, the status of an MNCRS replica is closer to that of a Lotus Notes client [Kaw92] which is aware only of the replicas that have been registered directly with it, and with which it may synchronize ....

[Article contains additional citation context not shown here]

Davidson, Susan B., Garcia-Molina, Hector, and Skeen, Dale. Consistency in partitioned networks. ACM Computing Surveys 17, No. 3 (September 1985), pp. 341-370

....conservative solutions [7, 25, 6, 13] These systems are largely inappropriate for general mobile computing requirements. The reference titled Consistency in a Partitioned Network: A Survey , contains a good survey on these systems, as well as coverage of several forms of optimistic replication [3]. 19 The growth of mobile computers has led many of the major database manufacturers to supply some form of replicated database solution that would allow some functionality to users working from portable machines. By and large, these systems support client server optimistic replication, with ....

S.B. Davidson; H. Garcia-Molina; D. Skeen, "Consistency in partitioned networks," Computing Surveys, vol.17, (no.3), p.341-70, Sept 1986.

....intersect [5] A more refined approach divides the quorum system into a collection of read quorums and a collection of write quorums such that any read quorum intersects any write quorum. Such systems have been used to implement distributed mutual exclusion [5] and data replication protocols [4, 7]. Quorums can be used with replicated data in transaction style synchronization that limits concurrency (cf. 2] whereas our goal and the goal of [1] is to reduce restrictions on asynchrony and concurrency. In this paper we present a service that emulates shared memory registers using ....

S.B. Davidson, H. Garcia-Molina and D. Skeen, "Consistency in Partitioned Networks", ACM Computing Surveys, vol. 15, no. 3, pp. 341-370, 1985.

No context found.

S. B. Davidson, H. Garcia-Molina, and D. Skeen, "Consistency in Partitioned Networks," ACM Comp. Surveys, vol. 17, no. 3, Sept., 1985.

No context found.

S. B. Davidson, H. Garcia-Molina, and D. Skeen. Consistency in partitioned networks. ACM Computing Surveys, 17(3), September 1985.

No context found.

S. Davidson, H. Garcia-Molina, and D. Skeen. Consistency in partitioned networks. Computing Surveys, 17(3):341--370, 1985.

No context found.

Davidson, S.B., Garcia-Molina, H., Skeen, D.: Consistency in partitioned networks. ACM Computing Surveys 17 (1985) 341--370

No context found.

S.B. Davidson, H. Garcia-Molina and D. Skeen, "Consistency in Partitioned Networks", ACM Computing Surveys, vol. 15, no. 3, pp. 341-370, 1985.

No context found.

Susan B. Davidson, Hector Garcia-Molina: "Consistency in Partitioned Networks", Computing Surveys 17(3), pp. 341-370, 1985.

No context found.

S. B. Davidson, H. Garcia-Molina, and D. Skeen. Consistency in partitioned networks. ACM Computing Surveys, 17(3), September 1985.

No context found.

S. B. Davidson, H. Garcia-Molina, D. Skeen, Consistency in partitioned networks, ACM Computing Surveys, vol 3, 341-370, 1985.

First 50 documents  Next 50

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