M.J. Fischer and A. Michael, "Sacrificing Serializability to Attain High Availability of Data in an Unreliable Network," ACM SIGACTSIGMOD Symposium on Principles of Database Systems, 1982, pp. 70--75.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....such as Orca [3] allows operations to be re ordered as long as they remain consistent with the view of individual clients. An inherent disparity in the performance of atomic and sequentially consistent objects has been established [2] Other systems provide even weaker guarantees to the clients [6, 5, 7] in order to get better performance. Improving performance by providing weaker guarantees results in more complicated semantics. Even when the behavior of the replicated objects is specified unambiguously, it is more difficult to understand and to reason about the correctness of implementations. ....

M. Fischer and A. Michael. Sacrificing serializability to attain high availability of data in an unreliable network. In Proceedings o] the A CM Symposium on Database Systems, pages 70-75, Mar. 1982.

....such as Orca [3] allows operations to be reordered as long as they remain consistent with the view of isolated clients. An inherent disparity in the performance of atomic and sequentially consistent objects has been established [2] Other systems provide even weaker guarantees to the clients [9, 5, 10] in order to get better performance. Improving performance by providing weaker guarantees results in more complicated semantics. Even when the behavior of the replicated objects is specified unambiguously, it is more difficult to understand and to reason about the correctness of implementations. ....

M. Fischer and A. Michael. Sacrificing serializability to attain high availability of data in an unreliable network. In Proceedings of the ACM Symposium on Database Systems, pages 70--75, Mar. 1982.

....messages on its own. The local competitiveness definition can be generalized using any positive valued tawet fuctio that measures the quality of the output. Approaches similar to local competitiveness were used in the past. For exampleFsee the best effort algorithm of Fischer and Michael [9] for database management. It may be interesting to note that the algorithm in [9] uses synchronized clocks. Some other work was done by Ajtai et al. 2]Fafter our preliminary paper was published [29] Loosely speakingF in [2] they consider a shared memory systemFwhere an execution is a sequence ....

....using any positive valued tawet fuctio that measures the quality of the output. Approaches similar to local competitiveness were used in the past. For exampleFsee the best effort algorithm of Fischer and Michael [9] for database management. It may be interesting to note that the algorithm in [9] uses synchronized clocks. Some other work was done by Ajtai et al. 2]Fafter our preliminary paper was published [29] Loosely speakingF in [2] they consider a shared memory systemFwhere an execution is a sequence of processor accesses to the shared memory. The order by which processors take ....

M. J. Fischer and A. Michael. Sacrificing serializability to attain high availability of data in an unreliable network. In Proc. ACM SIGACT-SIGMOD Symposium on Principles of Database SystemsFpages 70 75F1982.

....interacts. 6. Related Work Our work builds on numerous previous results in the area of highly available distributed systems and algorithms, including general replication techniques such as voting [11, 13, 2] and the primary copy method [1, 27, 26] Our work is also related to gossip schemes [12, 10, 34]. In this section, we focus on the most closely related work, namely providing high availability for applications where operations need not be ordered identically at all replicas. In this light, we compare our method with the relevant gossip schemes and with the work on ISIS. We also consider ....

....for applications where operations need not be ordered identically at all replicas. In this light, we compare our method with the relevant gossip schemes and with the work on ISIS. We also consider approaches where consistency is relaxed in order to improve performance. Some gossip methods [10, 34] require that a replica of the service exist at every client node, leading to increased 26 storage requirements and message traffic. Furthermore, these methods are application specific and do not allow the operation order to be controlled by clients. In the replication method used in the ....

Fischer, M. J., and Michael, A. Sacrificing Serializability to Attain High Availability of Dam in an Unreliable Network. In Proc. of the Symposium on Principles of Database Systems, pages 70-75. ACM, March, 1982.

....object associated with a given sync ID, the capabilities provided by iterating through sync entries could have been provided instead by a method returning an iterator over all sync IDs corresponding to objects in the sync store. The StoreManager class uses the abstract factory design pattern [Gam95] to create new sync stores. The framework includes an interface named SyncStoreFactory. This interface declares a single method, which attempts to create a SyncStore object consistent with a set of attributes specified in the method call. Each class implementing SyncStoreUpdater is accompanied by ....

.... pure local version, corresponding to a local update with a given update counter value and no remote updates methods to translate between various representations of the same version value Each of these methods returns a SyncVersionImpl result, allowing us to apply the Abstract Factory pattern of [Gam95] once again. Rather than declaring these methods in SyncVersionImpl, we declare them in an abstract class named SyncVersionFactory. Indeed, the method to return the earliest possible version and the method to return a pure local version are not naturally associated with an existing version ....

Fischer, Michael J., and Michael, Alan. Sacrificing serializability to attain high availability of data in an unreliable network. Proceedings of the ACM Symposium on Principles of Database Systems, March 29-31, 1982, Los Angeles, California, pp. 70-75

....For increased network utilization, Bengal packs as many records as possible into a single network packet. Eventually all updates must be propagated to all replicas. However, this requirement does not imply that all replicas must directly reconcile with each other, as shown in Figure 2. Gossiping [5] allows replica A to send changes to replica B, which sends its own and A s changes to replica C. Replica C might never communicate directly with replica A, but nonetheless C receives all updates applied to A s version of the data through replica B. Insert Figure 2a here Insert Figure 2b here ....

M.J. Fischer and A. Michael. "Sacrificing Serializability To Attain High Availability of Data In an Unreliable Network," Proceedings of the 4 th Symposium on Principles of Database Systems, 1982.

....order of updates to that object. Sequential consistency would prevent inconsistencies that could occur in cache consistency when multiple objects were being updated. For some create modify delete applications causal consistency is su#cient for correctness. One example is a distributed dictionary [20, 46]. A dictionary is replicated on many processes. Any process can insert, delete, or look up words in the dictionary. Di#erent processes can insert or delete the same word concurrently. Imagine a process receives two inserts and a delete for the same word. Was this word inserted, deleted, and ....

M. J. Fischer and A. Michael. Sacrificing serializability to attain high availability of data in an unreliable network. In Proceedings of the ACM Symposium on Principles of Database Systems, pages 70--75, March 1982.

....messages on its own. The local competitiveness definition can be generalized using any positive valued target function that measures the quality of the output. Approaches similar to local competitiveness were used in the past. For example, see the best effort algorithm of Fischer and Michael [9] for database management. It may be interesting to note that the algorithm in [9] uses synchronized clocks. Some other work was done by Ajtai et al. 2] after our preliminary paper was published [29] Loosely speaking, in [2] they consider a shared memory system, where an execution is a ....

....any positive valued target function that measures the quality of the output. Approaches similar to local competitiveness were used in the past. For example, see the best effort algorithm of Fischer and Michael [9] for database management. It may be interesting to note that the algorithm in [9] uses synchronized clocks. Some other work was done by Ajtai et al. 2] after our preliminary paper was published [29] Loosely speaking, in [2] they consider a shared memory system, where an execution is a sequence of processor accesses to the shared memory. The order by which processors take ....

M. J. Fischer and A. Michael. Sacrificing serializability to attain high availability of data in an unreliable network. In Proc. ACM SIGACT-SIGMOD Symposium on Principles of Database Systems, pages 70--75, 1982.

....one replica, and another, that is serviced by all the replicas. For example, a transaction processing system ensuring 1 copy serializability is one such application, and lazyreplication is another. In the former the updates are propagated synchronously, and in the latter they are asynchronous. In [Fischer and Michael 1992] there is a proposal of another 2 type protocol, namely one that maintains a dictionary database using three type of operations. Two update operations, INSERT and DELETE, and a query operation, LIST. In principle, the performance of any 2 type application can be improved by dynamic replication. ....

Fischer, M. and Michael, A. 1992. Sacrificing serializability to attain high availability of data in an unreliable network. In ACM Principles of Database Systems, pp. 70--75.

....same way by all processes; put another way, two concurrent write operations can be perceived in distinct orders by different processes. Several implementations of CC in various contexts and for different types of applications (such as CSCW and distributed dictionary management) have been explored [1, 5, 8, 16]. Systems that support ordered group communication provide users with an easy way to implement CC. For example, a communication primitive such as the ISIS cbcast primitive [4] which ensures all messages are delivered according to causal order) can be used in the following way. Each write ....

....size of the control information carried by this message is O(jY j n) The set Y has been defined at line (4.1) in Section 3. 5) In the worst case, this size is O(m n) and is similar to the one encountered in protocols that ensure causal delivery of all the messages exchanged during a computation [4, 5, 12, 16]. This price seems to be inherent to causality. Reduction of the size of the control information requires to add more synchronization but then we get a stronger consistency criterion 4 . So, the advantage of our protocol over classical causal delivery protocols lies in the possibility to discard ....

M.J. Fischer and A. Michael. Sacrificing Serializability to Attain High Availability of Data in an Unreliable Network. In Proc. ACM Symposium on Principles of Data Base Systems, 1982, pp. 70-75.

....in this paper we explore a weaker consistency criterion for distributed services, namely causal consistency. This consistency criterion takes into account only causality relations among accesses to objects. It appears that it has two main advantages: 1) it is meaningful for several applications [1, 5, 11, 22], and (2) protocols that implement it require only weak synchronization that permits the design of efficient implementations. One aim of this paper is the design of such a protocol for causally consistent distributed services. We consider a distributed model in which three sets of nodes are ....

....We will see in Section 4.1 how this dependence can be tracked in the case of causal consistency. PI n1039 10 M. Raynal 4. 1 Tracking Causality The most fundamental data structure to track causality relations and implement causally consistent objects is a two dimensional matrix of integers [5, 22, 3, 17]. Although a version vector is sufficient for tracking causality in the full replication case that was discussed previously, partial replication requires that updates to objects be recorded separately. For example, consider server node s i with copies of objects x a and x b , and server node s j ....

M.J. Fischer and A. Michael. Sacrificing Serializability to Attain High Availability of Data in an Unreliable Network. In Proc. ACM Symposium on Principles of Data Base Systems, 1982, pp. 70-75. Irisa An Adaptive Architecture for Causally Consistent Distributed Services 19

....a newly created object, or does the missing replica represent a recently deleted object Attempting to determine whether an object is newly created or recently deleted is futile in the absence of additional information. This create delete ambiguity (first noted by Fischer and Michael in [3]) is resolved in conventional replication schemes by appealing to an authority; in optimistic replication, some other means must be used. In this paper, we provide solutions for this and other problems typically encountered in optimistic replication. 1.1 File systems The algorithms presented in ....

....more limited replica management algorithms, from which the algorithms presented here are descended. The Coda project [11] has similar goals to our own Ficus work and bases its replica management on the LOCUS version vector [9] mechanism and an earlier draft of this work [4] Fischer and Michael [3] proposed recasting the replicated directory maintenance problem as a replicated dictionary problem, with slightly (but significantly) different semantics. A timestamp vector was used to infer from a comparison of two dictionary replicas which entries had been inserted and which had been ....

Michael J. Fischer and Alan Michael. Sacrificing serializability to attain high availability of data in an unreliable network. In Proceedings of the ACM Symposium on Principles of Database Systems, March 1982.

....may needlessly revoked for the sake of efficient analysis. Optimistic non serializable approaches have been proposed for computer conferencing [Str81] commercial information retrieval services [ABG87] bulletin board [BJS86] databases [BK85, SKS86, GAB83, All83, Fai81] and file directories [FM82, PPR83, Guy87]. Most of these base correctness criteria upon the semantics of the data and the operations performed. 1.5.2.4 Replicated directory management The directory replication problem has received particular attention because of the central role it plays is designing a highly reliable distributed file ....

....The directory replication problem has received particular attention because of the central role it plays is designing a highly reliable distributed file system. A wide range of directory replication mechanisms have been proposed, from serializable to non serializable. Fischer and Michaels [FM82] presented the first detailed examination of directory replication. They recast the problem as a replicated dictionary problem, to focus on the basic insert and delete operations common to each. Unsynchronized, concurrent directory modifications are resolved via timestamps. Various inefficiencies ....

[Article contains additional citation context not shown here]

Michael J. Fischer and Alan Michael. "Sacrificing Serializability to Attain High Availability of Data in an Unreliable Network." In Proceedings of the ACM Symposium on Principles of Database Systems, March 1982.

.... has recently been described by Koutsoupias and Papadimitriou [41] In addition, there is a long history of interest in optimality of a distributed algorithm given certain conditions, such as a particular pattern of failures [26, 31, 37, 43, 44, 45] or a particular pattern of message delivery [12, 33, 47]. In a sense, work on optimality envisions a fundamentally different role for the adversary in which it is trying to produce bad performance both in the candidate algorithm and in what we would call the champion algorithm; in contrast, the adversary used in competitive analysis usually cooperates ....

M. Fischer and A. Michael, Sacrificing serializability to attain high availability of data in an unreliable network. Research Report 221, Yale U., Feb. 1982.

....this cost. This gives rise to different notions of consistency. Sequential consistency [11] guaranteed by systems such as Orca [12] allows operations to be reordered as long as they remain consistent with the view of isolated clients. Other systems provide even weaker guarantees to the clients [13, 14, 15] to get better performance. Fekete et al. 1] defined a highly available eventually serializable data service (ESDS) They specified general conditions for such a service, and presented an algorithm based on lazy replication, in which operations received by each replica are gossiped in the ....

M. Fischer and A. Michael. Sacrificing Serializability to Attain High Availability of Data in an Unreliable Network. In Proceedings of the ACM Symposium on Database Systems, pp. 70-75, March 1982. 17

....This gives rise to different notions of consistency. Sequential consistency [11] guaranteed by systems such as Orca [12] allows operations to be reordered as long as they remain consistent with the view of isolated clients. Other systems provide even weaker guarantees to the clients [13, 14, 15] to get better performance. Improving performance by providing weaker consistency guarantees may lead to more complicated semantics. While in practice, replicated systems are often incompletely or ambiguously specified, it remains very important to provide formal consistency guarantees. Ladin, ....

M. Fischer and A. Michael. Sacrificing Serializability to Attain High Availability of Data in an Unreliable Network. In Proceedings of the ACM Symposium on Database Systems, pp. 70-75, March 1982.

....server directories are maintained redundantly on multiple sites and updates are incorporated in a lazy This research was partially supported by LANL under grant number 6863V0016 3A, by CALTRANS under grant number 65V250A, and by the NSF under grant numbers CDA94 21978 and CCR95 05807. manner [12, 33] through gossip messages [21, 17] epidemic propagation, and anti entropy [11] In this paper we use the epidemic communication model as the basis for a fault tolerant algorithm that supports transaction processing in replicated databases. In an epidemic system, sites perform update operations and ....

M. J. Fischer and A. Michael. Sacrificing Serializability to Attain High Availability of Data in an Unreliable Network. In Proceedings of the First ACM Symposium on Principles of Database Systems, pages 70--75, May 1982.

....have been performed. Their results were recently extended by Prisco, Mayer and Yung [49] There is a long history of interest in optimality of a distributed algorithm given certain conditions, such as a particular pattern of failures [25, 29, 34, 46] or a particular pattern of message delivery [12, 31, 48]. These and related works are in the spirit of our paper, but differ substantially in the details and applicability to distinct situations. 3 Model of Computation We assume a system of n processes p 1 ; pn , that communicate through shared memory. Each location in memory is called a ....

M. Fischer and A. Michael, Sacrificing Serializability to Attain High Availablity of Data in an Unreliable Network. Research Report 221, Yale U., Feb. 1982.

....their right to decide on abort or commit to the commit coordinator. Network partitions or site failures may thus lead to smaller availability of the database system. Not all applications should pay the price for maintaining a consistent view on replicated data and immediate update propagation [16], 20] 30] Examples of such applications are automated teller machine networks, flight reservation, and part inventory control. For instance, it is clearly unacceptable for a flight reservation system to become globally unavailable in case of a site failure or network partition. It is ....

....a preliminary description of how to apply history logs for propagating independent updates was given by us in [10] Update propagation was considered also in the context of maintaining replicated dictionaries, with an approach which has several similarities with the one proposed in this paper. In [16], a vector is introduced in order to keep track, at each site, of the 4 S.CERI, M.A.W. HOUTSMA, A.M. KELLER, P. SAMARATI events that originated in the system. However, only the insert and delete operations are allowed, and the vector does not give any indication of how up to date other sites of ....

[Article contains additional citation context not shown here]

M.J. Fischer and A. Michael, "Sacrificing Serializability to Attain High Availability of Data in an Unreliable Network," ACM SIGACTSIGMOD Symposium on Principles of Database Systems, 1982, pp. 70--75.

....UGC CUHK Research Grants. cope with these situations. 1.1 Related Work Research work on accessing data from the remote file server and data availability can be traced back to the period of designing distributed file systems. Both the theoretical and the architectural aspect can be found in [13, 14]. Davidson S. B. 17] provided a comprehensive discussion on the tradeoff between data availability and correctness. Based on the optimistic replicate control protocol [15] experimental mobile file system like the Coda [16] and the Ficus [11, 12] were built. Both the Coda and the Ficus allow ....

Fischer, M. J., and Michael, A., Sacrificing serializability to attain high availability of data in an unreliable network, Proceedings of the 1st ACM SIGACT-SIGMOD Symposium on Principles of Database Systems, May., In ACM, New York, pp. 70-75. 1982.

....that are distinct from clients and as a result provides better performance: it requires many fewer messages than the process group approach, the messages are smaller, and it can tolerate network partitions. Our technique is based on the gossip approach first introduced by Fischer and Michael [9], and later enhanced by Wuu and Bernstein [36] and our own earlier work [16] We have extended the earlier work in two important ways: by supporting causal ordering for updates as well as queries and by incorporating forced and immediate operations to make a more generally applicable method. The ....

....counter, and the initial (zero) timestamp contains zero in each part. Timestamps are partially ordered in the obvious way: t s ( t s . t s ) 1 1 n n Two timestamps t and s are merged by taking their component wise maximum. Multipart timestamps were used in Locus [30] and also in [9, 13, 16, 22, 36]. Both uids and labels are represented by multipart timestamps. Every update operation is assigned a unique multipart timestamp as its uid. A label is created by merging timestamps; a label timestamp t identifies the updates whose timestamps are less than or equal to t. The dependency relation is ....

[Article contains additional citation context not shown here]

Fischer, M. J., and Michael, A. Sacrificing Serializability to Attain High Availability of Data in an Unreliable Network. Proc. of the Symposium on Principles of Database Systems, ACM, March, 1982, pp. 70-75.

.... has recently been described by Koutsoupias and Papadimitriou [40] In addition, there is a long history of interest in optimality of a distributed algorithm given certain conditions, such as a particular pattern of failures [26, 30, 36, 42, 43, 44] or a particular pattern of message delivery [12, 32, 46]. In a sense, work on optimality envisions a fundamentally different role for the adversary in which it is trying to produce bad performance both in the candidate algorithm and in what we would call the champion algorithm; in contrast, the adversary used in competitive analysis usually cooperates ....

M. Fischer and A. Michael, Sacrificing serializability to attain high availability of data in an unreliable network. Research Report 221, Yale U., Feb. 1982.

No context found.

M.J. Fischer and A. Michael, "Sacrificing Serializability to Attain High Availability of Data in an Unreliable Network," ACM SIGACTSIGMOD Symposium on Principles of Database Systems, 1982, pp. 70--75.

No context found.

Fischer, M. J. and Michael, A. (1982) Sacrificing serializability to attain high availability of data in an unreliable network. In Proc. 1st ACM Ann. Symp. on Principles of Database Systems, pp. 70--75.

No context found.

Michael J. Fischer and A. Michael. Sacrificing serializability to attain high availability of data. In Proceedings of the ACM Symposium on Principles of Database Systems, pages 70--75. ACM, 1982.

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