P. Chundi, D. J. Rosenkrantz, and S. S. Ravi. Deferred Updates and Data Placement in Distributed Databases. In Proc. of the 12th Int. Conf. on Data Engineering, pp. 469--476, 1996.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....correctness but it does greatly simplify our implementations. Without such an assumption, a proper UNDO mechanism will be required to maintain replica consistency. 5. 2 Replica Consistency and Failure Recovery In general, data replication is achieved through either eager or lazy write propagations [24, 44]. Eager propagation keeps all replicas exactly synchronized by acquiring locks and updating data at all replicas in a globally coordinated manner. In comparison, lazy propagation allows lock acquisitions and data updates to be completed independently at each replica. Previous work shows that ....

....network services in which service data can be divided into independent partitions. Therefore, Neptune s consistency model does not address the data consistency across partition boundaries. Neptune s first two levels of replica consistency are more or less generalized from the previous work [24, 67] and we provide an extension in the third level to address the data staleness problem from two different perspectives. Notice that a consistency level is specified for each service and thus Neptune allows co existence of services with different consistency levels. Level 1. Write anywhere ....

[Article contains additional citation context not shown here]

P. Chundi, D. J. Rosenkrantz, and S. S. Ravi. Deferred Updates and Data Placement in Distributed Databases. In Proc. of the 12th Intl. Conf. on Data Engineering, pages 469--476, New Orleans, Louisiana, February 1996.

....Optimizations. 1 Introduction The replicated data management in distributed databases is a classic problem with great practical importance. Distributed data warehouses and data marts contain a huge amount of replicated data distributed among a number of sites. Therefore, in recent developments [3, 4, 6, 10] of the area there is always a trade o# among system e#ciency, data availability, data freshness, and data consistency. Two replicated data management methods are available in the literature: eager and lazy. Eager replication management gives the data consistency and the highest data freshness. ....

P. Chundi, D.J. Rosenkratz, and S.S. Ravi, Deferred Updates and Data Placement in Distributed Databases, Proceedings of the Twelveth International Conference on Data Engineering, New Orleans, Louisiana, 1996.

....provider and aggregator (left half of the figure) Directory servers maintain the mapping between human understandable hierarchical names to object GUIDs. The hierarchical tree can be easily stored in a relational database and various partitioning [2] or primary secondary replication schemes [3, 9, 33] can be applied to provide scalability and fault tolerance. a storage aggregator object locator mcast subscriber provider table addr info ip slave bloom filter storage availability system load an entry in the provider table provider p value a storage provider catalog manager ....

.... a single cluster node in Figure 1 (shadowed area marked with B) which may be preferable for clusters with functionally symmetric nodes [31] Data redundancy, consistency, concurrency control and conflict resolution have been extensively studied in storage systems [26, 27] distributed databases [9, 13, 30, 33] and distributed file systems [4, 19, 24] and are not the focus of this work. We briefly describe the schemes adopted in Sorrento on these three aspects as follows: 1) Data redundancy. Sorrento uses a combination of replication and erasure coding [23, 27] to survive data through software ....

CHUNDI,P.,ROSENKRANTZ, D., AND RAV , S. S. Deferred updates and data placement in distributed databases. In ICDE (1996), pp. 469--476.

....Optimizations. 1 Introduction The replicated data management in distributed databases is a classic problem with great practical importance. Distributed data warehouses and data marts contain a huge amount of replicated data distributed among a number of sites. Therefore, in recent developments [3, 4, 6, 10] of the area there is always a trade o among system eciency, data availability, data freshness, and data consistency. Two replicated data management methods are available in the literature: eager and lazy. Eager replication management gives the data consistency and the highest data freshness. ....

P. Chundi, D.J. Rosenkratz, and S.S. Ravi, Deferred Updates and Data Placement in Distributed Databases, Proceedings of the Twelveth International Conference on Data Engineering, New Orleans, Louisiana, 1996.

....each service module provides a CHECK callback so that the Neptune server module can check if a previously spawned service instance has been successfully completed. 3 Replica Consistency and Failure Recovery In general, data replication is achieved through either eager or lazy write propagations [5, 9]. Eager propagation keeps all replicas exactly synchronized by acquiring locks and updating data at all replicas in a globally coordinated manner. In comparison, lazy propagation allows lock acquisitions and data updates to be completed independently at each replica. Previous work shows that ....

....current version of Neptune does not have full fledged transactional support largely because Neptune restricts each service access to a single data partition. 3. 1 Multi level Consistency Model Neptune s first two levels of replica consistency are more or less generalized from the previous work [5, 17] and we provide an extension in the third level to address the data staleness problem from two different perspectives. Notice that a consistency level is specified for each service and thus Neptune allows co existence of services with different consistency levels. Level 1. Write anywhere ....

[Article contains additional citation context not shown here]

P. Chundi, D. J. Rosenkrantz, and S. S. Ravi. Deferred Updates and Data Placement in Distributed Databases. In Proc. of the 12th Intl. Conf. on Data Engineering, pages 469--476, New Orleans, Louisiana, February 1996.

.... change because other transactions may have affected the balance) and unacceptable for others (an item is out of stock and cannot be sold) This technique does a good job in solving a fundamental issue: standard ways of propagating updates to replicas (eager replication [26] 25] lazy replication [21], 56] make deadlocks increase as the cube of the number of sites and as the fourth power of transaction size, thereby rendering the systems unscalable. On the other hand, the two tier technique may result in an unacceptable number of failed transactions (after the clients reconnect) and user ....

# P. Chundi, D.J. Rosenkratz, and S.S. Ravi, "Deferred Updates and Data Placement in Distributed Databases," Proc. 12th Int'l Conf. Data Eng., New Orleans, 1996.

....use of a lazy approach to update of secondary copies of replicated data and the use of a new concept, virtual sites, to reduce the potential for conflict among global transactions. 1 Introduction The problem of consistent access to replicated data has re emerged as a challenge in recent years [CRR96, GHOS96, HHB96, PL91, SAB 96] with the advent of distributed data warehouses and data marts at the high end, and distributed data in often disconnected mobile computers at the low end [KI96] The fundamental problem, as identified by [GHOS96] is that the standard transactional approach to ....

....constraint that the sum of the balances must be positive. 2 To avoid anomalies such as the one illustrated above and nevertheless guarantee global serializability, the lazy master approach must be augmented with one of the following: ffl Restrictions on how primary copies of data are selected[CRR96] ffl A global concurrency control mechanism that minimizes coordination among sites. In this paper, we choose the latter approach and present an alternative approach to lazy propagation that has fewer deadlocks than the approach of [GHOS96] permits local read only transactions to run without ....

[Article contains additional citation context not shown here]

P. Chundi, D. J. Rosenkrantz, and S. S. Ravi. Deferred updates and data placement in distributed databases. In Proceedings of the Twelveth International Conference on Data Engineering, New Orleans, Louisiana, 1996.

....object, which eases the protocol design. The single coordinator property ensures that no conflicting updates will occur even in case of network partitions. It is worth noting that the replication model adopted in island based file system is similar to the models in typical replicated databases [17] [18] However, our consistency requirement differs, primarily because we do not require transactional semantics for file accesses. Therefore, we believe that an approach to the consistency of island based file system will be valuable to other file systems in this model as well. 6. Consistency ....

P. Chundi, D. J. Rosenkratz, and S. S. Ravi, "Deferred Updates and Data Placement in Distributed Databases", in Proceedings of 12 th International Conference on Data Engineering, 1996.

....update propagation can eliminate these deficiencies. Recently, much attention has been directed towards the lazy approach to replica update propagation. The lazy approach requires that installation of updates to replicas occur only after the update transaction has committed at the origination site [8, 10, 13, 16, 17, 5, 1]. Propagation is performed by independent subtransactions spawned only after the transaction at the origination site has committed. Clearly, the activities of these subtransactions must be managed carefully to ensure global consistency and transaction atomicity. In [8] the authors introduced the ....

....site [8, 10, 13, 16, 17, 5, 1] Propagation is performed by independent subtransactions spawned only after the transaction at the origination site has committed. Clearly, the activities of these subtransactions must be managed carefully to ensure global consistency and transaction atomicity. In [8] the authors introduced the data placement graph, and proved that replica consistency can be guaranteed by ensuring the acyclicity of that graph. However, their approach does not guarantee global serializability in the general case. It requires that each local DBMS use rigorous two phase locking ....

P. Chundi, D. J. Rosenkrantz, and S. S. Ravi. Deferred updates and data placement in distributed databases. In Proceedings of the Twelveth International Conference on Data Engineering, New Orleans, Louisiana, 1996.

....Base Endowment. To copy otherwise, or to republish, requires a fee and or special permission from the Endowment. Proceedings of the 26th VLDB Conference, Cairo, Egypt, 2000. et al. GHOS96] and, since the publication of those results, the research focus has shifted towards lazy replication [CRR96, PMS99, ABKW98, BKR 99] The drawback of lazy replication is that, if consistency is necessary, many non trivial problems arise. Namely, in the case of update everywhere (each copy can be updated) maintaining consistency is usually left to the user. If only a primary copy can be updated, ....

....the research side, lazy replication has been studied using very different approaches like weak consistency models [PL91, KB91, GN95] economic paradigms [SAS 96] or epidemic strategies [AES97] More recent work has explored lazy strategies that still provide consistency. Thus, Chundi et al. CRR96] have shown that in lazy primary copy schemes, serializability cannot be guaranteed without restricting the placement of primary and secondary copies in the system. Recent work by Pacitti et al. PMS99] and Breitbart et al. BKR 99] has attempted to minimize this limitation. As a major ....

P. Chundi, D. J. Rosenkrantz, and S. S. Ravi. Deferred updates and data placement in distributed databases. In Proc. of the Int. Conf. on Data Engineering, 1996.

....update protocols to eliminate all requirements on data placement. The extension is a hybrid protocol that propagates as many updates as possible in a lazy fashion. Before discussing our specific contributions in more detail, we present the system model adopted in this paper and prior work from [CRR96, GHOS96, BK97, ABKW98] 1.1 System Model The model we adopt in this paper is very similar to one from [CRR96, BK97] For each data item, a particular site is chosen as its primary site. The copy of a data item at the primary site is called the primary copy and the other copies are referred to ....

....as many updates as possible in a lazy fashion. Before discussing our specific contributions in more detail, we present the system model adopted in this paper and prior work from [CRR96, GHOS96, BK97, ABKW98] 1. 1 System Model The model we adopt in this paper is very similar to one from [CRR96, BK97] For each data item, a particular site is chosen as its primary site. The copy of a data item at the primary site is called the primary copy and the other copies are referred to as secondary copies or replicas. Each transaction originates at a single site and is a sequence of read and ....

[Article contains additional citation context not shown here]

P. Chundi, D. J. Rosenkratz, and S. S. Ravi. Deferred updates and data placement in distributed databases. In Proceedings of the Twelveth International Conference on Data Engineering, New Orleans, Louisiana, 1996.

....update protocols to eliminate all requirements on data placement. The extension is a hybrid protocol that propagates as many updates as possible in a lazy fashion. Before discussing our specific contributions in more detail, we present the system model adopted in this paper and prior work from [CRR96, GHOS96, BK97, ABKW98]. 1.1 System Model The model we adopt in this paper is very similar to one from [CRR96, BK97] For each data item, a particular site is chosen as its primary site. The copy of a data item at the primary site is called the primary copy and the other copies are referred to as secondary copies or ....

....as many updates as possible in a lazy fashion. Before discussing our specific contributions in more detail, we present the system model adopted in this paper and prior work from [CRR96, GHOS96, BK97, ABKW98] 1. 1 System Model The model we adopt in this paper is very similar to one from [CRR96, BK97]. For each data item, a particular site is chosen as its primary site. The copy of a data item at the primary site is called the primary copy and the other copies are referred to as secondary copies or replicas. Each transaction originates at a single site and is a sequence of read and write ....

[Article contains additional citation context not shown here]

P. Chundi, D. J. Rosenkratz, and S. S. Ravi. Deferred updates and data placement in distributed databases. In Proceedings of the Twelveth International Conference on Data Engineering, New Orleans, Louisiana, 1996.

....only one copy of a replicated data item. After the transaction commits, the update is propagated asynchronously to the other copies. Reference [Go94] discusses several methods used by commercial database systems to implement deferred update. A classification of these approaches is provided in [CRR96]. The vendors of commercial database systems that provide deferred update [Ib94, Mo94, Co93, MP 93, Or93] discuss several applications for which a deferred update protocol seems preferable to two phase commit. However, they do not address the issue of serializability in those systems. In fact, ....

....The vendors of commercial database systems that provide deferred update [Ib94, Mo94, Co93, MP 93, Or93] discuss several applications for which a deferred update protocol seems preferable to two phase commit. However, they do not address the issue of serializability in those systems. In fact, [CRR96] provides examples of non serializable scenarios in systems with a deferred update capability. The paper Things Every Update Replication Customer Should Know [Go95] also discusses the importance of serializability in a deferred update system and concludes that Asynchronous update replication ....

[Article contains additional citation context not shown here]

P. Chundi, D. J. Rosenkrantz, and S. S. Ravi, "Deferred Updates and Data Placement in Distributed Databases", To appear in ICDE '96, New Orleans, LA, Feb 1996.

No context found.

P. Chundi, D. J. Rosenkrantz, and S. S. Ravi. Deferred Updates and Data Placement in Distributed Databases. In Proc. of the 12th Int. Conf. on Data Engineering, pp. 469--476, 1996.

No context found.

P. Chundi, D. J. Rosenkrantz, and S. S. Ravi. Deferred updates and data placement in distributed databases. In IEEE Int. Conf. on Data Engineering, pages 469--476, 1996.

No context found.

P. Chundi, D. J. Rosenkrantz, and S. S. Ravi. Deferred updates and data placement in distributed databases. In IEEE Int. Conf. on Data Engineering, pages 469--476, 1996.

No context found.

P. Chundi, D. J. Rosenkrantz, and S. S. Ravi. Deferred updates and data placement in distributed databases. In IEEE Int. Conf. on Data Engineering, pages 469--476, 1996.

No context found.

Chundi, P., Rosenkrantz, D.J., Ravi, S.S.: Deferred updates and data placement in distributed databases. In: Proceedings of the 12th International Conference on Data Engineering, IEEE Computer Society (1996) 469--476

No context found.

Parvathi Chundi, Daniel J. Rosenkrantz, and S. S. Ravi. Deferred updates and data placement in distributed databases. In Proceedings of the 12th International Conference on Data Engineering, pages 469--476. IEEE Computer Society, February 1996.

No context found.

Chundi, P., Rosenkrantz, D.J., Ravi, S.S.: Deferred updates and data placement in distributed databases. In: Proceedings of the 12th International Conference on Data Engineering, IEEE Computer Society (1996) 469--476

No context found.

Parvathi Chundi, Daniel J. Rosenkrantz, and S. S. Ravi. Deferred updates and data placement in distributed databases. In Proceedings of the 12th International Conference on Data Engineering, pages 469--476. IEEE Computer Society, February 1996.

No context found.

Parvathi Chundi, Daniel J. Rosenkrantz, and S. S. Ravi. Deferred updates and data placement in distributed databases. In Proceedings of the 12th International Conference on Data Engineering, pages 469--476. IEEE Computer Society, February 1996.

No context found.

P. Chundi, D.J. Rosenkrantz, and S.S. Ravi, "Deferred Updates and Data Placement in Distributed Databases," Proc. 12th Int'l Conf. Data Eng., pp. 469-476, Feb. 1996.

No context found.

P. Chundi, D. J. Rosenkrantz, and S. S. Ravi. Deferred updates and data placement in distributed databases. In Proc. of the Int. Conf. on Data Engineering (ICDE), pages 469-476, New Orleans, Louisiana, February 1996.

No context found.

P. Chundi, D. J. Rosenkrantz and S. S. Ravi. Deferred Updates and Data Placement in Distributed Databases. In International Conference on Data Engineering, Louisiana, February 1996

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