E. N. Hanson. A performance analysis of view materialization strategies. In SIGMOD, 1987.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....of data. 27] introduced the concept of continuous queries but is restricted to append only databases. The cache of an LMR can be viewed as a set of materialized views with the corresponding rules as view definitions. 15, 4, 21] present some algorithms for updating materialized views. In [17] di#erent strategies to update materialized views are investigated with respect to their performance. There are also similarities with semantic caching as described in [12] 6 Conclusion and Future Work In this paper, we presented the architecture of MDV, a distributed metadata management ....

E. Hanson. A Performance Analysis of View Materialization Strategies. In Proc. of the ACM SIGMOD Conf.onManagementofData, pages 440--453, 1987.

....sniffing. This process identifies (a) a mapping between cached results and the corresponding queries used to generate those results and (b) a mapping between the queries and the data changes that affect these queries. The second problem is closely related to the prob lem of view maintenance [4, 5, 6, 7] in the context of materialized views in data warehouses. Since a data warehouse consists of a large view, the main fo cus of the database research has been to maintain materialized views incrementally. Numerous algorithms have been proposed for incremental view main tenance [8, 9, 10, 11, 12] ....

E. N. Hanson. A Performance Analysis of View Materialization Strategies. In A CM pages 440-453, May 1987.

....only when a query is issued against the view, and periodic update [LHM 86] which updates the view at periodic intervals. Performance studies on these strategies have determined that the efficiency of an approach depends heavily on the structure of the base relations and on update patterns [Han87] We assume immediate update in this paper, but we observe that with little or no modifi our algorithms can be applied to deferred and periodic update as well. The algorithms in this paper could be viewed as a specialized concurrency control mechanism for a multidatabase system. Our ....

E.N. Hanson. A performance analysis of view materialization strategies. In Proceedings of the ACM SIGMOD International Conference on Management of Data, pages 440--453, 1987.

....to provide sample views in his (her) DBMS, it is not self evident that one should support the maintenance of materialized sample views. The rationale is the traditional one: materialized views may be more efficient than virtual views (query modification) for supporting queries against the views [Han87b] In general, materialized views are to be preferred when queries against the views are common, and updates of the underlying relations relatively infrequent [Han87b] For sample views, the case for materialized views is even stronger, since sampling fractions from very large databases are ....

.... one: materialized views may be more efficient than virtual views (query modification) for supporting queries against the views [Han87b] In general, materialized views are to be preferred when queries against the views are common, and updates of the underlying relations relatively infrequent [Han87b] For sample views, the case for materialized views is even stronger, since sampling fractions from very large databases are typically quite small, a few percent or less. Thus most sample queries are effectively highly selective queries. Since most updates will therefore not alter the sample ....

[Article contains additional citation context not shown here]

Eric N. Hanson. A Performance Analysis of View Materializations Strategies. In ACM SIGMOD International Conference on the Management of Data, pages 440--453, May 1987.

....serving stale data can have catastrophic consequences. The view selection problem aims at balancing the trade off between performance improvement and maintenance overhead because of materialization. View selection has been studiedextensively in relational databases and data warehouses [Rou82, Han87, Sel88, BM90, GM95, RCK 95, SSV96, Gup97, CD97, KR99] In data warehouses, View Selection is performed off line during the down time of the warehouse. Web servers, on the other hand, must remain online all the time and thus, updates are applied in the back end database while the web server ....

Eric N. Hanson. "A Performance Analysis of View Materialization Strategies". In Proc. of the ACM SIGMOD Conference, pages 440--453, San Francisco, California, May 1987.

....are more likely to happen [HS93] Different issues concerning the caching technique have also been studied. AL80, Rou82b, Val87, Rou91] proposed alternative methods for storing the cached data, Sel88, Jhi88] discussed the problem of selective caching and cache replacement, and in [RK86, Han87, BLT86] different strategies for updating cached data are explored. Aside from the above work which focused on the problem of cache maintenance and management, the problem of how to identify the useful cached data for computing queries, referred as query matching problem, was addressed in ....

....via re execution, the technique of incremental update (or differential computation) LHM86, BLT86, Rou91] can efficiently update a temporary if only a little part of it is changed. For the second update method, logs must be maintained to support differential computations. It was analyzed in [Han87] that none of the combinations of update strategy and update method is superior to all the others under all situations. As it is practically prohibitive to experiment with all possible combinations, on demand strategy has been adopted in our implementation because it can batch consecutive updates ....

E.N. Hanson. A performance analysis of view materialization strategies. In Procs. of ACM-SIGMOD, pages 440--453, 1987.

....view. 5 Conclusions The recent increase in research on identifying which views should be materialized indicates the importance of studies into the performance of incremental maintenance techniques. The most comprehensive study we can nd of the performance of IVM is that of Hanson [8], which is an analytical model of immediate and deferred update propagation in relational databases. We are not aware of any research that has sought to validate the models presented by Hanson experimentally, or that has systematically assessed the performance of view maintenance techniques in ....

Eric N. Hanson. A Performance Analysis of View Materialization Strategies. In ACM Press, editor, Proc. SIGMOD, pages 440-453, 1987.

....in a simplistic way, can be seen as a generalization of view maintenance used in relational databases. A DW maintenance policy determines when and how to refresh a view to reflect changes to its sources. Various view maintenance policies have been suggested and analysed in the literature [Bla86, Han87, Seg89, Seg91]. It has been shown that the selection of the most appropriate maintenance policy is a complex process affected by several dynamic and static system properties [Han87, Sri88, Col97] The autonomy, distribution and 2 heterogeneity aspects of data warehousing have motivated additional studies to ....

....changes to its sources. Various view maintenance policies have been suggested and analysed in the literature [Bla86, Han87, Seg89, Seg91] It has been shown that the selection of the most appropriate maintenance policy is a complex process affected by several dynamic and static system properties [Han87, Sri88, Col97]. The autonomy, distribution and 2 heterogeneity aspects of data warehousing have motivated additional studies to address the new dimensions of the maintenance problem. There have been several approaches [Agr97, Hul96, Zhu99] to incrementally maintaining views and preserving consistency when ....

[Article contains additional citation context not shown here]

E.N. Hanson, "A Performance Analysis of View Materialization Strategies", SIGMOD Conference, 1987

....assume that the select project operations are performed according to the de nition of the updating view. 3 Diculties in Maintaining a Hierarchy of Views in a Consistent State The advantages of maintaining views incrementally rather than through recomputation has been recognized in previous work [6, 7]. Several algorithms have been proposed to ensure consistent updates at a view. However, they are mostly tailored for the case of a single view incorporating changes from single or multiple data sources. In the case when views are allowed to derive updates from other views, there is a need for ....

E. N. Hanson. A Performance Analysis of View Materialization Strategies. In Proceedings of the ACM SIGMOD International Conference on Management of Data, pages 440-453, May 1987.

....pages occupied by the input relations is n then, in most situations, our algorithm allows an O(n) evaluation cost, thereby improving on the O(n 2 ) cost of nested loop evaluation and the O(nlog(n) cost of sort merge evaluation. We then adapt the algorithm to an incremental mode of computation [Han87, Rou87, Hor86] where the join is materialized and updated incrementally as the base relations change. We motivate why incremental evaluation is an appropriate strategy for valid time join evaluation. The paper is organized as follows. Section 2 formally defines the valid time natural join in ....

....of long lived tuples. 5 Incremental Evaluation Incremental query evaluation attempts to reduce the amount of recomputation performed by the DBMS. In this strategy, a query is materialized as a view, and once computed, it is stored back in the database for later use [Rou87, BCL86, SRMF87, Han87] When an update occurs to a base relation, differentials are propagated to the views that depend on that base relation, incrementally ensuring the consistency of the view. Applications of incremental query evaluation have been found in the areas of smart editing [HT86] mainframe workstation ....

[Article contains additional citation context not shown here]

E. N. Hanson. A Performance Analysis of View Materialization Strategies. In Proceedings of ACM SIGMOD, pages 440--453, San Francisco, CA, May 1987.

....have to be considered, and unless c is kept small, the effort involved in simply bounding the search could become significant. 6. 3 Incremental Changes Whenever some derived information is materialized, a change in the base information needs to be reflected in a change in the derived information [9, 15, 18, 28]. We require precomputed shortest distances between domain centers, and between each domain center and its constituent nodes. Whenever a modification is made to the original graph, this precomputed information has to be updated. Obviously, a complete recomputation would be extremely expensive. One ....

E. Hanson, "A Performance Analysis of View Materialization Strategies," Proc. ACM-SIGMOD 1987 Int'l Conf. on Management of Data, San Fransisco, California, May 1987, pp. 440-453.

.... query modification [Sto75] A pointerbased ViewCache contains a set of tuple identifiers that identify and point to the data objects in the underlying relations that be dereferenced to materialize the view [Rou91] A materialized view is stored as the set of data objects that belong to the view [Han87] ViewCaches and materialized views must be maintained incrementally. We present and analyze five refreshment algorithms that are based on various combinations of materialized views, partially materialized views, and ViewCaches. Existing data warehousing approaches focus solely on materialized ....

.... [BM95b] View indexing by means of pointer structures has been described in [Baek95] BM95a] BM95b] BM95c] JMR91] QW91] Rou82a] Rou82b] Rou91] RK91] Val87] The use and refreshment of materialized views has been de2 scribed in [BM90] CKPS95] GL95] GMR95] HK95] Han87] LMSS95] SJGP90] SR87] ZGMHW95] Incremental techniques has been described for active databases [CSL 90] CW91] HCKW90] Han92] RCBB89] WDSY91] for deductive databases [GMS93] and for temporal databases [BM95c] JMR91] McK88] However, we are not aware of existing ....

E.N. Hanson. A Performance Analysis of View Materialization Strategies. In Proceedings of the 1987 ACM SIGMOD International Conference on Management of Data, San Fransisco, California, pages 440--453, 1987.

....the subsequent executions of a given CQ query is to use differential evaluation method such that queries that can be answered using delta information (i.e. the updated data) rather than the full set of base data. Similar techniques have been widely used in incremental view materialization [5, 14, 19, 16, 34]. More concretely, recall Chapter 2, we have defined a CQ query as a sequence of query results, modeled by Q cq (S 1 ) Q cq (Sn ) where Q cq (S i ) i = 1; n) is the result of running Q cq on the database state S i . Quite often there are situations where users are more interested in ....

....insertion, deletion or modification (see Example 4) We define an operator that computes differential relations for arbitrary SPJ expressions. The concept of differential relations is, to some extent, similar to the concept of hypothetical relations used for incremental updating materialized views[5, 14]. The difference lies in the usage and the detailed structure. In the eager mechanism for materialized view update, a hypothetical relation represents the net changes made by a single transaction to a base relation and can be dropped after the transaction is committed and the materialized view is ....

[Article contains additional citation context not shown here]

E. N. Hanson. A performance analysis of view materialization strategies. In Proceedings of the ACM-SIGMOD International Conference on Management of Data, pages 440--453, San Francisco, CA, May 1987.

....studied outside the mainstream of deductive database research. Maybe the most important of these applications is the maintenance of materialized views, i.e. of rule defined relations the extension of which is 1 explicitly stored at all times in order to avoid costly derivations at query time [Han87]. Both integrity checking and refreshment of materialized views are special cases of monitoring conditions over changing databases. So called monitors and alerters have been suggested in the past [BuCle79] General triggering facilities are re gaining particular interest at the moment in ....

E. Hanson: "A Performance Analysis of View Materialization Strategies", Proc. 13th ACM-SIGMOD Conf. on Management of Data, San Francisco, May 1987

....immediate or deferred. In an immediate timing policy [2, 3] a materialized view is maintained within or immediately after the transaction that changes the source relations. In a deferred timing policy [5] the maintenance of the view is delayed. It can be done periodically [34] at query time [16], or on demand by the user. It can also be triggered by events at the sources (e.g. when the net change to a source database exceeds a certain threshold) Sometimes supporting multiple maintenance timing policies may be appropriate [6, 33] Type of environment. Most of the research work on ....

E. Hanson. A Performance Analysis of View Materialization Strategy. In Proc. of the ACM SIGMOD Intl. Conf. on Management of Data, pages 440--453, 1987.

....the view name with a subquery) or by materializing (instantiating and maintaining) the view. View materialization is typically preferable when the views are heavily used and the update traffic is light. For a discussion of when view materialization is preferable to query modification see Hanson (1987), Segev and Park (1989) Segev and Fang (1990) Segev and Fang (1991) The basic idea is to reuse the maximal portion of the original sample when constructing the updated sample. Our results are thus a synthesis of classical techniques for updating materialized views Shmueli and Itai (1984) ....

....idea is to reuse the maximal portion of the original sample when constructing the updated sample. Our results are thus a synthesis of classical techniques for updating materialized views Shmueli and Itai (1984) Shmueli and Itai (1987) Blakeley, Larson and Tompa (1986) Tompa and Blakely (1988) Hanson (1987), and Ceri and Widom (1991) with classical sampling algorithms. Devroye (1991) discusses a similar idea of sample reuse (coupled samples) in a simulation setting. Sample reuse is the basis of panel surveys which are widely used for longitudinal studies. Below we treat only the case of maintaining ....

Hanson, E. N. (1987). A performance analysis of view materializations strategies, ACM SIGMOD International Conference on the Management of Data, pp. 440--453.

....of connectivity. Systems such as in [27, 82] divide updates to the server into those immediately affecting the materialized view which must be sent to the client immediately and those updates which may be deferred. Some systems only update views when related queries are executed against the views[78, 136]. All of these systems include the need for on demand connectivity; therefore, they are not appropriate for our environment. Similar concepts have been tried in file systems for intermittent connectivity such a CODA[96] To increase scalability, we will perform data partitioning by vertical and ....

Eric R. Hanson. A performance analysis of view materialization strategies. In Umeshwar Dayal and Irv Traiger, editors, Proceedings of ACM SIGMOD '87, volume 16, pages 440--453, May 1987. May 27-29, San Fransico, CA.

....Within the traditional context, the second case occurs if an update to a base table occurs which possibly affects a materialized view. The resulting problem preserving the consistency of the view is called view maintenance and has been discussed for, e.g. for (extended) relation 2 [8, 23, 38] and deductive databases [30, 24, 21, 44, 37] The same problem occurs also for the materialization of functions within object bases [28] if the values of some object s attributes change, the materialized function value becomes invalid. However, since we do not necessarily materialize the view ....

....at time t. Likewise, Deltaf Gamma t;t 1 ( args ) is the set of objects returned by executing function f at time t that are not returned when f is executed at time t 1. Note that the efficient computation of the difference between two successive database states has been extensively studied [8, 28, 30, 23, 38]. However, as we will see, we do not need the difference explicitly for our view maintenance mechanism. We only use it to investigate the effects of an update to an external function onto a materialized mediated view if T P is used. For a constraint atom to be introduced into the materialized ....

E. Hanson. A performance analysis of view materialization strategies. In Proc. of the ACM SIGMOD Conf. on Management of Data, pages 440--453, 1987.

....cost computation, we assumed a uniform distribution for the transaction size Z i . The default parameter values shown in Table 10 were used in the cost computation unless stated otherwise. The parameter values such as the processing speeds c io was chosen similarly as in other literature, e.g. (Hanson, 1987). Figures 5 illustrates the expected cost of rule activation per unit time as a function of join dimension (n) The chart shows that S2, the First Join Pattern Relation 0 20 40 60 80 100 120 140 160 180 2 3 4 5 6 7 8 9 10 11 join dimension, n expected cost per unit time (sec 1000sec) S1 S2 S3 ....

E. R. Hanson. (1987). A performance analysis of view materialization strategies. In Proceedings of ACM SIGMOD International Conference on the Management of Data, May.

....servers and network. Updates can be applied to data at the server side meaning a primary copy paradigm only. Aging of cached data is allowed to avoid complicated algorithms ensuring cache coherency at the clients. The cached data items are called quasi copies and are similar to materialized views [22] and database snapshots [1,27] Generally, caching promises reduced network and server load as well. In the quasi copy approach the degree of coherency relaxation has to be specified at the client side by means of coherency conditions like e.g. times t 0 : k : 0 k a) x (t) x(t k) ....

E. Hanson: "A Performance Analysis of View Materialization Strategies", SIGMOD 87, pages 440-453.

....functions have been investigated in [KMWZ91] The functions are used for deriving attribute values which are stored in a separate data structure. A similar strategy is pursued in the extended relational database system Postgres [SK91] Cost models and benchmark results are in [KKM91] and [Han87] for relational databases) Object oriented Query Optimization. Object oriented database systems like O 2 [OT92, BCD92] and ObjectStore [OHMS92] focus query optimization on the use of physical clustering strategies and indexes. In O 2 , indexes are quite flexible by allowing a membership ....

E.N. Hanson. A performance analysis of view materialization strategies. In Proc. 1987 ACM SIGMOD Int. Conf. on Management of Data, pages 440--453, San Francisco, CA, May 1987.

....algorithms appears in [GM95] Materialized views were investigated in the 1980 s as a tool to speed up queries on views. Several view maintenance algorithms were proposed [NY83, Pai84, SI84, BLT86, LHM 86, BCL89, SP89, QW91] analytical models for the cost of view maintenance were developed [Han87, SR88, SF90] and the impact of materialized views on the performance of queries was studied [Han87, BM90] The use of view maintenance algorithms in maintaining integrity constraints was also recognized [BC79] However, implementations of materialized views in research prototypes and commercial ....

....up queries on views. Several view maintenance algorithms were proposed [NY83, Pai84, SI84, BLT86, LHM 86, BCL89, SP89, QW91] analytical models for the cost of view maintenance were developed [Han87, SR88, SF90] and the impact of materialized views on the performance of queries was studied [Han87, BM90] The use of view maintenance algorithms in maintaining integrity constraints was also recognized [BC79] However, implementations of materialized views in research prototypes and commercial systems were missing (with the sole exception of ADMS [Rou91] The benefit of materialized views to ....

Eric N. Hanson. A performance analysis of view materialization strategies. In Umeshwar Dayal and Irv Traiger, editors, Proceedings of ACM SIGMOD 1987 International Conference on Management of Data, pages 440--453, San Francisco, CA, May 27-29 1987.

....only when a query is issued against the view, and periodic update [LHM 86] which updates the view at periodic intervals. Performance studies on these strategies have determined that the efficiency of an approach depends heavily on the structure of the base relations and on update patterns [Han87] We assume immediate update in this paper, but we observe that with little or no modification our algorithms can be applied to deferred and periodic update as well. The algorithms in this paper could be viewed as a specialized concurrency control mechanism for a multidatabase system. Our ....

E.N. Hanson. A performance analysis of view materialization strategies. In Proceedings of the ACM SIGMOD International Conference on Management of Data, pages 440--453, 1987.

....However, it is well known that joins are the most time consuming operations. By focusing on materializing joins, we are able to achieve significant time reductions in query evaluation, as shown in our experiments on a real application. Materializing views has been studied in various contexts [13, 24]. View selection was recently studied in [12] that takes, as input, a fixed set of queries, their evaluation plans and costs. Often in applications, queries are fixed and and it is even more difficult to obtain the evaluation plans and costs. The selection problem was also considered in OLAP ....

E. N. Hanson. A performance analysis of view materialization strategies. In Proc. ACM SIGMOD Int. Conf. on Management of Data, 1987.

.... is being externally updated can be thought of as a materialized views of some external databases (in our example, the databases at the New York Stock Exchange, the Tokyo Exchange, and so on) Although the question of how to efficiently export a materialized view has been studied in great detail [Han87, LHM 86, RK86, SP89, AL80, BLT86, SG90] the question of how to efficiently import a view has not been studied at all. It is usually assumed, often implicitly, that the problem of incorporating view updates is straightforward and that the resource demand it represents is trivial. We believe ....

E. Hanson. A performance analysis of view materialization strategies. In Proceedings of the ACM SIGMOD Annual Conference on Management of Data, pages 440--453, 1987.

....can now be discussed in light of the example. We start by first noting that the system should support several concurrent modeling efforts. This means to allow multiple users to access the same set of data, which is essentially the same as supporting multiple views of the underlying storage system [BCL86, BLT86, LHM86, Han87]. Consider for instance the two uses made of the soil map. The erosion model is likely to focus more on certain properties of the terrain such as its physical composition. The vegetation model, on the other hand, probably uses more the chemical properties of the terrain such as its ph (in both ....

E.N. Hanson. A Performance Analysis of View Materialization Strategies. Proc. of the SIGMOD International Conference on Management of Data, San Francisco, CA, pages 440--453, May 1987.

....volume and weight computed there is no way to speed up the evaluation with an index. On the other hand, if volume and weight were ordinary attributes, we could provide indices to enhance efficiency. Function materialization, analogous to view materialization in relational database systems [10, 11], makes it possible to retrieve the precomputed result with much lesser cost, exactly like any other attribute [12] It is also possible to define indices on the materialized results to support queries like the one above. Of course, there are additional efforts required to keep the materialized ....

E. Hanson, "A performance analysis of view materialization strategies," in Proc. of the ACM SIGMOD Conf. on Management of Data, (San Francisco, CA), pp. 440--453, May 87.

....associated with query shipping vs. data shipping and caching (e.g. DR93] closely parallel the trade offs in the mediator architecture between the virtual and materialized approaches. Furthermore, the use of materialization in supporting views in a conventional DBMS architecture (e.g. Han87] has similarities to the use of materialization in mediators. A key difference between the mediator context and the other contexts is that with mediators, the source databases are presumed to be largely autonomous from each other and from the mediator. In particular, mediated environments ....

E. N. Hanson. A performance analysis of view materialization strategies. In Proc. ACM SIGMOD Symp. on the Management of Data, pages 440--453, 1987.

....technique with some minor improvements optimizations can be translated into an optimal incremental maintenance algorithm under the above cost model. 8 Related Work A large body of work exists describing different algorithms for incrementally maintaining materialized views [BLT86, RK86, Han87, BCL89, CW91, QW91, GMS93, GLT94, GL95, CGL 96, GJM97, Qua97, GK98] Each work applies to different classes of views and has various advantages and disadvantages. Below, we discuss some of the above algorithms. Qian and Wiederhold in [QW91] present a technique (later corrected in [GLT94] to ....

E. Hanson. A performance analysis of view materialization strategies. In Umeshwar Dayal and Irv Traiger, editors, Proceedings of ACM SIGMOD 1987 International Conference on Management of Data, pages 440--453, San Francisco, CA, May 27-29 1987.

....the subsequent executions of a given CQ query is to use differential evaluation method such that queries that can be answered using delta information (i.e. the updated data) rather than the full set of base data. Similar techniques have been widely used in incremental view materialization [3, 5, 7, 6, 13]. More concretely, recall Sect. 2, we have defined a CQ query as a sequence of query results, modeled by Q cq (S 1 ) Q cq (Sn ) where Q cq (S i ) i = 1; n) is the result of running Q cq on the database state S i . Quite often there are situations where users are more interested in ....

E. N. Hanson. A performance analysis of view materialization strategies. In Proceedings of the ACM-SIGMOD International Conference on Management of Data, pages 440--453, San Francisco, CA, May 1987.

....the insertions to the views in V based on 4T . 7 Related Work One of the problems that our framework tackles is how to maintain a view when only parts of the underlying tables are accessible. Most work on view maintenance assumes that the complete underlying tables are accessible, for example, [BLT86,CW91,GL95,GMS93,Han87,QW91]. However, there has also been work on view maintenance that assumes otherwise. BT88] and [GJM96] identified self maintainable views that can be maintained without accessing underlying tables. QGMW96] HZ96] and [Qua97] tried to make a view self maintainable by defining auxiliary views such ....

E. Hanson. A performance analysis of view materialization strategies. In Proceedings of the 1987 ACM SIGMOD International Conference on Management of Data, pages 440--453, May 1987.

.... [Lai:88] 2 Multi database schemas to integrate data [Smith:81] Dayal:83] 3 Methods to achieve closure [Beeri:87] Ramakrishnan:89] 4 Library catalogs and indexing aids [Humphrey:87] 5 Thesaurus structures for access to bibliographic systems [McCarthy:88] 6 Materialized views and view indexes [Hanson:87] [Roussopoulos:86] 7 Rule bases for semantic query optimization [King:84] Chakravarthy:85] 8 Object generators [Wiederhold:86] Barsalou:88] 9 Knowledge bases to access multiple databases [Sacca:86] 10 Computations on derived data [Hammer:78] Adiba:81] 11 Graphic tools to help in hypothesis ....

Eric Hanson: "A Performance Analysis of View Materialization Strategies "; Proc. ACM-SIGMOD 87, May 1987.

....re issue their queries frequently to obtain the new data that match their queries. Trying to monitor the constant information flow in the Internet is both labor intensive for the user and resource intensive for the system. Unfortunately, traditional database techniques such as view materialization [3, 7], active databases [20] and query optimization, while powerful by themselves, encounter difficulties in distributed interoperable environments such as the Internet. The interoperation of these database techniques with each other and with the relatively unorganized storage and management of data, ....

....base table [3] The second defers the refresh until a query is issued against the view [19] The third refreshes the view periodically [9] The main tradeoff in choosing among these approaches is the staleness of the view data vs. the cost of updating it. Most of the algorithms in the literature [3, 9, 7, 8, 6] work in a centralized database environment, in which the materialized view and its base tables co reside. The study on distributed materialized view management has been primarily focused on determining the optimal refresh sources and timing for multiple views defined on the same base data [22, ....

[Article contains additional citation context not shown here]

E. N. Hanson. A performance analysis of view materialization strategies. In Proceedings of the ACMSIGMOD International Conference on Management of Data, pages 440--453, San Francisco, CA, May 1987.

....is acyclic, and the history is valid. Corollary 9 states that an execution history is Mat Serializable if its MatSerialization graph for view level conflicts is acyclic and if each transaction sees the effects of its own writes. 4 Related Work View Maintenance: Most work on view maintenance [2, 4, 7, 10, 11, 12, 16] has dealt with automatic derivation of the maintain function from the view definition. 9] is a survey and classification of several view maintenance algorithms. 5] deals with deriving the maintain functions specifically for deferred maintenance. Our current work is different from all of the ....

E. N. Hanson. A performance analysis of view materialization strategies. In U. Dayal and I. Traiger, editors, Proceedings of ACM SIGMOD 1987 International Conference, pages 440--453, San Francisco, CA, May 1987.

....extended relational models, in particular the POSTGRES data model [21] In POSTGRES an attribute of type POSTQUEL may consist of a relational query that has to be evaluated upon referencing. Much work has been spent on optimizing the evaluation of queries accessing such POSTQUEL attributes, e.g. [6, 7, 9, 18, 19, 20], by caching the results. The above cited work is similar to ours with respect to the general idea of precomputing results. However, exploiting object oriented features facilitates a much finer grained control over rematerialization requirements of precomputed results in our approach than is ....

E. Hanson. A performance analysis of view materialization strategies. In Proc. of the ACM SIGMOD Conf. on Management of Data, pages 440--453, San Francisco, CA, May 87.

....and rematerialization overhead upon update operations. In this respect function materialization is related to relational view materialization [1, 3, 4] Analogous work exists for the POSTGRES data model to optimize the evaluation of queries accessing (virtual) POSTQUEL attributes, e.g. [6, 7, 8, 13, 14, 15]. The above cited work is similar to ours with respect to the general idea of precomputing (caching) results. However, the exploitation of object oriented features, especially the classification of objects into types , object identity , and the principle of encapsulation facilitates a much ....

E. Hanson. A performance analysis of view materialization strategies. In Proc. of the ACM SIGMOD Conf., pages 440--453, San Francisco, CA, May 87.

....only when a query is issued against the view, and periodic update [LHM 86] which updates the view at periodic intervals. Performance studies on these strategies have determined that the efficiency of an approach depends heavily on the structure of the base relations and on update patterns [Han87] We assume immediate update in this paper, but we observe that with little or no modification our algorithms can be applied to deferred and periodic update as well. 3 Correctness Our first task is to define what correctness means in an environment where activity at the source is decoupled ....

E.N. Hanson. A performance analysis of view materialization strategies. In Proceedings of the ACM SIGMOD International Conference on Management of Data, pages 440--453, 1987.

....it meets the needs of specific applications. Typically, the contents of a view are derived using the view s stored query, in which case each query upon the extent of a view must be translated into a query upon the view s source classes. In relational systems, this is known as query modification [4]. View materialization the maintenance of derived extents of views as opposed to recomputing them upon demand has been demonstrated in the relational model to improve query performance. If the view is materialized, then its contents must be maintained in the face of updates to the view s ....

E. N. Hanson. A performance analysis of view materialization strategies. SIGMOD, pages 440--453, 1987.

....interoperability by hiding idiosyncrasies of component systems to be integrated into one unified, yet federated system. View materialization, i.e. the maintenance of derived data instead of recomputing it on demand, has been demonstrated in the relational model to improve query access times [7]. Possible applications of materialized view technology include providing selective replication for partitioned or centralized distributed systems, and aiding with the integration of federated DBMSs by providing replication transparency. This work was supported in part by the NSF RIA grant ....

E. N. Hanson. A performance analysis of view materialization strategies. SIGMOD, pages 440-- 453, 1987.

....doing this. 5 Related Work One of the problems that our framework tackles is how to maintain a view when only parts of the underlying views are available. Most of the work on view maintenance have assumed that the complete underlying views are available ( BLT86] BT88] CW91] GL95] GMS93] Han87] QW91] and many others) However, VO 4 VO 1.4 DW 2 1 VOM Record input SM User Interface CreateVO CreateAuxVO DoDML DoDML spawn spawn 5 3 4 Figure 9: View Definition Object Interaction DW VO 4 SM 1 DoDML 3 4 CompensateA CompensateA AnswerQuery 2 VO 1 DoDML 5 Figure 10: View Control ....

E. Hanson. A performance analysis of view materialization strategies. In Proceedings of 1987 ACM SIGMOD, pages 440--453, 1987.

....to define several such summary tables with one statement. As changes are made to the data sources, the warehouse views must be updated to reflect the changed state of the data sources. The views either can be recomputed from scratch, or incremental maintenance techniques [BC79, SI84, RK86, BLT86, Han87, SP89, QW91, Qua96, CW91, GMS93, GL95, LMSS95, ZGHW95] can be used to calculate the changes to the views due to the source changes. Using the incremental maintenance approach, the warehouse can be updated either immediately as soon as a change from a source is received, or the update can be ....

E. Hanson. A performance analysis of view materialization strategies. In Proceedings of ACM SIGMOD 1987 International Conference on Management of Data, pages 440--453, San Francisco, CA, May 1987.

....it meets the needs of specific applications. Typically, the contents of a view are derived using the view s stored query, in which case each query upon the extent of a view must be translated into a query upon the view s source classes. In relational systems, this is known as query modification [9]. View materialization, i.e. the maintenance of derived extents of views as opposed to recomputing them upon demand, has been demonstrated in the relational model to improve the performance of queries. If the view is materialized, then its contents must be maintained in the face of updates to ....

E. N. Hanson. A performance analysis of view materialization strategies. SIGMOD, pages 440--453, 1987.

....to delay recomputations slightly, so that several related base updates can be combined in a single recomputation. The challenge is to accomplish this without severely affecting the timeliness of the derived data. There are a number of studies related to the derived data maintenance problem (e.g. [BLT86, BCL86,Han87,AL80,RK86,SF90]) Most of these studies assume the relational model and focus on maintaining materialized views expressible by relational algebra and some basic aggregate functions such as sum and average . While materialized views can be considered as a special case of derived data, our study covers the more ....

....performing the base data updates. We believe that decoupled recomputations are best for a high performance system, so in this paper we only consider this type of recomputation. 3. Incremental Recomputations. Recomputations can be performed more efficiently if they are done incrementally [Han87]. Instead of fully recomputing a derived item by reading the values of all of the base data items it depends on, it is recomputed using its old value and the set of changes to the base data since its last recomputation. Significant performance gain, due to fewer data accesses and shorter ....

[Article contains additional citation context not shown here]

E. Hanson. A performance analysis of view materialization strategies. In Proceedings of the ACM SIGMOD Annual Conference on Management of Data, pages 440--453, 1987.

.... materialized view (the action) Following a more complex yet more efficient approach, updates with no effect to the view are not propagated (which is checked in condition parts of ECA rules) and entire recomputations of the view are avoided by propagating only incremental changes to the view (cf. [3, 9]) Example 3. Rule propagateNetPriceUpdates of class ITEM propagates any update of netPrice to the derived and stored variable price (cf. Fig. 1) Note: The if clause of an ECA rule may be omitted if the condition is the predicate true. 2.3 Maintaining Transitional Constraints Transitional ....

E. Hanson. A Performance Analysis of View Materialization Strategies. In Proceedings of the ACM SIGMOD Conference on Management of Data, pages 440--453, May 1987.

No context found.

E. N. Hanson. A performance analysis of view materialization strategies. In SIGMOD, 1987.

No context found.

E. N. Hanson. A performance analysis of view materialization strategies. In Proceedings of the ACM-SIGMOD International Conference on Management of Data, pages 440#453, San Francisco, CA, May 1987.

No context found.

Hanson, E. N., "A performance analysis of view materialization strategies," in Proceedings of the ACM-SIGMOD International Conference on Management of Data, (San Francisco, CA), pp. 440--453, May 1987.

No context found.

E.N. Hanson, A performance analysis of view materialization strategies, in: Proc. SIGMOD, 1987, pp. 440--453.

No context found.

E.N. Hanson. A performance analysis of view materialization strategies. In Proc. of 5th ACM SIGMOD Intl. Conf. on Management of Data, pages 440 -- 453, New York, 1987. ACM Press.

No context found.

E. Hanson. (1987) A performance analysis of view materialization strategies., In Proc. of the ACM SIGMOD Conf. on Management of Data, pp. 440--453, 1987.

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