Replica Refresh Strategies in a Database Cluster

DMCA

Replica Refresh Strategies in a Database Cluster (2005)

Cached

Download Links

Other Repositories/Bibliography

DBLP

by Cecile Le Pape , Stephane Gancarski , Patrick Valduriez

Venue:	in "Journées Bases de Données Avancées (BDA05
Citations:	1 - 0 self

Citations

217	Data caching issues in an information retrieval system. - Alonso, Barbara, et al. - 1990 (Show Context) Citation Context ... Thus, depending on the consistency/performance requirements, eager and lazy replication can be both useful in database clusters. Relaxing consistency using lazy replication has gained much attention =-=[1,3,19,30,26,16]-=-, even quite recently [11]. The main reason is that applications often tolerate to read data that is not perfectly consistent, which can be exploited to improve performance. However, replica divergenc...
185	Don’t be lazy, be consistent: Postgres-R, a new way to implement database replication. In: - Kemme, Alonso - 2000 (Show Context) Citation Context ...icas. By exploiting efficient group communication services provided by a cluster, eager replication can be made non blocking (unlike with distributedstransactions) and scale up to large cluster sizes =-=[13,14,24,12]-=-. Furthermore, it makes query load balancing easy. With lazy replication, a transaction updates only one replica and the other replicas are updated (refreshed) later on by separate refresh transaction...
133	A new approach to developing and implementing eager database replication protocols. - Kemme, Alonso - 2000 (Show Context) Citation Context ...icas. By exploiting efficient group communication services provided by a cluster, eager replication can be made non blocking (unlike with distributedstransactions) and scale up to large cluster sizes =-=[13,14,24,12]-=-. Furthermore, it makes query load balancing easy. With lazy replication, a transaction updates only one replica and the other replicas are updated (refreshed) later on by separate refresh transaction...
104	Offering a precision-performance tradeoff for aggregation queries over replicated data. In - Olston, Widom - 2000 (Show Context) Citation Context ... Thus, depending on the consistency/performance requirements, eager and lazy replication can be both useful in database clusters. Relaxing consistency using lazy replication has gained much attention =-=[1,3,19,30,26,16]-=-, even quite recently [11]. The main reason is that applications often tolerate to read data that is not perfectly consistent, which can be exploited to improve performance. However, replica divergenc...
100	Algorithms for Deferred View Maintenance - Colby, Griffin, et al. - 1996 (Show Context) Citation Context ...ular strategy is to propagate updates from the source to the copies as soon as possible (ASAP), as in [4,5,7]. Another simple strategy is to refresh replicas periodically [6,18] as in data warehouses =-=[8]-=-. Another strategy is to maintain the freshness level of replicas, by propagating updates only when a replica is too stale [28]. There are also mixed strategies. In [20], data sources push updates to ...
91	The demarcation protocol: A technique for maintaining constraints in distributed database systems. - Barbara, Garcia-Molina - 1994 (Show Context) Citation Context ... Thus, depending on the consistency/performance requirements, eager and lazy replication can be both useful in database clusters. Relaxing consistency using lazy replication has gained much attention =-=[1,3,19,30,26,16]-=-, even quite recently [11]. The main reason is that applications often tolerate to read data that is not perfectly consistent, which can be exploited to improve performance. However, replica divergenc...
63	Update propagation protocols for replicated databases. - Breitbart, Komondoor, et al. - 1999 (Show Context) Citation Context ... proposed in the context of distributed databases, data warehouse and and database clusters. A popular strategy is to propagate updates from the source to the copies as soon as possible (ASAP), as in =-=[4,5,7]-=-. Another simple strategy is to refresh replicas periodically [6,18] as in data warehouses [8]. Another strategy is to maintain the freshness level of replicas, by propagating updates only when a repl...
61	Scalable replication in database clusters. In: - Patino-Martınez, Jimenez-Peris, et al. - 2000 (Show Context) Citation Context ...icas. By exploiting efficient group communication services provided by a cluster, eager replication can be made non blocking (unlike with distributedstransactions) and scale up to large cluster sizes =-=[13,14,24,12]-=-. Furthermore, it makes query load balancing easy. With lazy replication, a transaction updates only one replica and the other replicas are updated (refreshed) later on by separate refresh transaction...
44	W.: A comparative evaluation of transparent scaling techniques for dynamic content servers. In: - Amza, Cox, et al. - 2005 (Show Context) Citation Context ...ent routing strategies, such as Random, (weighted) Round-robin, Shortest Queue first, Shortest execution first (SELF), and so on. The current version of Refresco implements a enhanced version of SELF =-=[2]-=- which includes the estimated cost of refreshing a node on-demand. It works as follows. When the router receives a transaction, it sends it to the master node. When the transaction has validated on th...
42	Efficient numerical error bounding for replicated network services. - Yu, Vahdat - 2000 (Show Context) Citation Context ... Thus, depending on the consistency/performance requirements, eager and lazy replication can be both useful in database clusters. Relaxing consistency using lazy replication has gained much attention =-=[1,3,19,30,26,16]-=-, even quite recently [11]. The main reason is that applications often tolerate to read data that is not perfectly consistent, which can be exploited to improve performance. However, replica divergenc...
40	Fas - a freshness-sensitive coordination middleware for a cluster of olap components. - Rohm, Bohm, et al. - 2002 (Show Context) Citation Context ...ing, refresh strategy. 1 Introduction Database clusters provide a cost-effective alternative to parallel database systems, i.e. database systems on tightly-coupled multiprocessors. A database cluster =-=[10,9,25,26]-=- is a cluster of PC servers, each running an off-the-shelf DBMS. A major difference with parallel database systems is the use of a “black-box” DBMS at each node. Since the DBMS source code is not nece...
38	Update propagation strategies to improve freshness in lazy master replicated databases. - Pacitti, Simon - 2000 (Show Context) Citation Context ...as been very few studies of refresh strategies and they are incomplete. For instance, they do not take into account the starting time of update propagation [27,15] or only consider variations of ASAP =-=[23]-=-. This paper has three main contributions. First, we propose a model which allows describing and analyzing existing refresh strategies, independent of other load balancing issues. Second, we describe ...
35	Relaxed Currency and Consistency: How to Say “Good Enough” in SQL. - Guo, Larson, et al. - 2004 (Show Context) Citation Context ...formance requirements, eager and lazy replication can be both useful in database clusters. Relaxing consistency using lazy replication has gained much attention [1,3,19,30,26,16], even quite recently =-=[11]-=-. The main reason is that applications often tolerate to read data that is not perfectly consistent, which can be exploited to improve performance. However, replica divergence must be controlled since...
34	Fast algorithms for maintaining replica consistency in lazy master replicated - Pacitti, Minet, et al. - 1999 (Show Context) Citation Context ...Furthermore, it makes query load balancing easy. With lazy replication, a transaction updates only one replica and the other replicas are updated (refreshed) later on by separate refresh transactions =-=[21,22]-=-. By relaxing consistency, lazy replication can better provide transaction load balancing, in addition to query load balancing. Thus, depending on the consistency/performance requirements, eager and l...
30	Deferred updates and data placement in distributed databases,” - Chundi, Rosenkrantz, et al. - 1996 (Show Context) Citation Context ... proposed in the context of distributed databases, data warehouse and and database clusters. A popular strategy is to propagate updates from the source to the copies as soon as possible (ASAP), as in =-=[4,5,7]-=-. Another simple strategy is to refresh replicas periodically [6,18] as in data warehouses [8]. Another strategy is to maintain the freshness level of replicas, by propagating updates only when a repl...
26	Balancing performance and data freshness in web database servers. - Labrinidis, Roussopoulos - 2003 (Show Context) Citation Context ... Thus, depending on the consistency/performance requirements, eager and lazy replication can be both useful in database clusters. Relaxing consistency using lazy replication has gained much attention =-=[1,3,19,30,26,16]-=-, even quite recently [11]. The main reason is that applications often tolerate to read data that is not perfectly consistent, which can be exploited to improve performance. However, replica divergenc...
23	Optimistic replication for internet data services - Saito, Levy - 2000 (Show Context) Citation Context ...nto account in the routing strategy. There has been very few studies of refresh strategies and they are incomplete. For instance, they do not take into account the starting time of update propagation =-=[27,15]-=- or only consider variations of ASAP [23]. This paper has three main contributions. First, we propose a model which allows describing and analyzing existing refresh strategies, independent of other lo...
22	Maintaining coherency of dynamic data in cooperative repositories. - Shah, Ramamritham, et al. - 1995 (Show Context) Citation Context ...strategy is to refresh replicas periodically [6,18] as in data warehouses [8]. Another strategy is to maintain the freshness level of replicas, by propagating updates only when a replica is too stale =-=[28]-=-. There are also mixed strategies. In [20], data sources push updates to cache nodes when their freshness is too low. However, cache nodes can also force refreshment if needed. In [16], an asynchronou...
20	Replica consistency in lazy master replicated databases - Pacitti, Minet, et al. - 2001 (Show Context) Citation Context ...Furthermore, it makes query load balancing easy. With lazy replication, a transaction updates only one replica and the other replicas are updated (refreshed) later on by separate refresh transactions =-=[21,22]-=-. By relaxing consistency, lazy replication can better provide transaction load balancing, in addition to query load balancing. Thus, depending on the consistency/performance requirements, eager and l...
16	Cache-aware query routing in a cluster of databases. - Rohm, Bohm, et al. - 2001 (Show Context) Citation Context ...ing, refresh strategy. 1 Introduction Database clusters provide a cost-effective alternative to parallel database systems, i.e. database systems on tightly-coupled multiprocessors. A database cluster =-=[10,9,25,26]-=- is a cluster of PC servers, each running an off-the-shelf DBMS. A major difference with parallel database systems is the use of a “black-box” DBMS at each node. Since the DBMS source code is not nece...
13	Parallel processing with autonomous databases in a cluster system. - Gancarski, Naacke, et al. - 2002 (Show Context) Citation Context ...ing, refresh strategy. 1 Introduction Database clusters provide a cost-effective alternative to parallel database systems, i.e. database systems on tightly-coupled multiprocessors. A database cluster =-=[10,9,25,26]-=- is a cluster of PC servers, each running an off-the-shelf DBMS. A major difference with parallel database systems is the use of a “black-box” DBMS at each node. Since the DBMS source code is not nece...
11	Standards for databases on the grid. - Malaika - 2003 (Show Context) Citation Context ...ood and could be used as default strategy. Our prototype allows the DBA to select the best strategy according to the workload type generated by the application. It is thus compliant with the OGSA-DAI =-=[18]-=- definition of a Data Resource Manager providing flexible and transparent access. The paper is organized as follows. Section 2 describes our database cluster architecture, with emphasis on load balanc...
10	An adaptive framework for tunable consistency and timeliness using replication,” - Krishnamurthy, Sanders, et al. - 2002 (Show Context) Citation Context ...nto account in the routing strategy. There has been very few studies of refresh strategies and they are incomplete. For instance, they do not take into account the starting time of update propagation =-=[27,15]-=- or only consider variations of ASAP [23]. This paper has three main contributions. First, we propose a model which allows describing and analyzing existing refresh strategies, independent of other lo...
10	The leganet system: freshness-aware transaction routing in a database cluster - Ganarski, Naacke, et al. - 2007 (Show Context) Citation Context ...fferent queries in the same workload, or we can vary the ratio query/transaction in a workload, and so on. We also plan to integrate the refresh strategies into the multi-master approach presented in =-=[9]-=-, as we suggested in [8]. The work presented in this paper can be seen as a first step toward a self-adaptable refresh strategy, which would combine different strategies by analysing on-line the incom...
9	Load balancing of autonomous applications and databases in a cluster system. - Gancarski, Naacke, et al. - 2002 (Show Context) Citation Context ...ing, refresh strategy. 1 Introduction Database clusters provide a cost-effective alternative to parallel database systems, i.e. database systems on tightly-coupled multiprocessors. A database cluster =-=[10,9,25,26]-=- is a cluster of PC servers, each running an off-the-shelf DBMS. A major difference with parallel database systems is the use of a “black-box” DBMS at each node. Since the DBMS source code is not nece...
6	Refresco: Improving Query Performance Through Freshness Control in a Database Cluster - Pape, Gançarski, et al. - 2004 (Show Context) Citation Context ...not perfectly consistent, and this can be exploited to improve performance. However, replica divergence must be controlled since refreshing replicas becomes more difficult as divergence increases. In =-=[16]-=-, we addressed this problem in the context of a shared-nothing database cluster. We chose mono-master lazy replication because it is both simple and sufficient in many applications, e.g. ASP, where mo...
4	Are quorums an alternative for database replication - Jiménez-Peris, Patino-Martinez, et al. - 2003 (Show Context) Citation Context ...icas. By exploiting efficient group communication services provided by a cluster, eager replication can be made non blocking (unlike with distributedstransactions) and scale up to large cluster sizes =-=[13,14,24,12]-=-. Furthermore, it makes query load balancing easy. With lazy replication, a transaction updates only one replica and the other replicas are updated (refreshed) later on by separate refresh transaction...
4	Refresco: Improving query performance through freshness control in a database cluster - Pape, Gançarski, et al. - 2004 (Show Context) Citation Context ...is not perfectly consistent, which can be exploited to improve performance. However, replica divergence must be controlled since refreshing replicas becomes more difficult as divergence increases. In =-=[17]-=-, we addressed this problem in the context of a shared-nothing database cluster. We chose mono-master lazy replication because it is both simple and sufficient in many applications, e.g. ASP, where mo...
3	A critique of ansi isolation levels - Berenson, Bernstein, et al. - 1995 (Show Context) Citation Context ... proposed in the context of distributed databases, data warehouse and and database clusters. A popular strategy is to propagate updates from the source to the copies as soon as possible (ASAP), as in =-=[4,5,7]-=-. Another simple strategy is to refresh replicas periodically [6,18] as in data warehouses [8]. Another strategy is to maintain the freshness level of replicas, by propagating updates only when a repl...
3	Scalable application aware data freshening - Carney, Lee, et al. - 2002 (Show Context) Citation Context ... and database clusters. A popular strategy is to propagate updates from the source to the copies as soon as possible (ASAP), as in [4,5,7]. Another simple strategy is to refresh replicas periodically =-=[6,18]-=- as in data warehouses [8]. Another strategy is to maintain the freshness level of replicas, by propagating updates only when a replica is too stale [28]. There are also mixed strategies. In [20], dat...
2	Scheduling queries to improve the freshness of a website - Liu, Ng, et al. - 2005
1	Fine-grained refresh strategies for managing replication in database clusters - Gançarski, Pape, et al. - 2005 (Show Context) Citation Context ...ame workload, or we can vary the ratio query/transaction in a workload, and so on. We also plan to integrate the refresh strategies into the multi-master approach presented in [9], as we suggested in =-=[8]-=-. The work presented in this paper can be seen as a first step toward a self-adaptable refresh strategy, which would combine different strategies by analysing on-line the incoming workload. According ...