D. F. Lieuwen and D. J. DeWitt, A Transformation-Based Approach to Optimizing Loops in Database Programming Languages, Proc. ACM SIGMOD Conf.,San Diego, CA, June 1992, 91.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....optimizer a complex and tedious task. Vo lcano salgebraic approach seems much more natural and easier to understand. Most recent work in object oriented query optimization and some work on database programming languages has focused on algebras and algebraic transformations, e.g. [LiD92, ShZ90, StO91, VaD91] among manyothers. Second, in order to avoid the problems associated with adding newoperators to the costbased optimizer,new operators are integrated at the query rewrite level. However, query optimization on the query rewrite levelisheuristic; in other words, it does not include cost estimation. ....

D. F. Lieuwen and D. J. DeWitt, A Transformation-Based Approach to Optimizing Loops in Database Programming Languages, Proc. ACM SIGMOD Conf.,San Diego, CA, June 1992, 91.

....the equivalences inv 1 5 is analogous. The approach of [Clu92] retains a separation of DML and programming language; thus, the possibility of non termination or infinite data structures are not considered. Several other groups have introduced object algebras and strategies for their optimisation [Dem94, Lie92, Sha89, Sto91, Van91] with analogous equivalences to those we propose here. In general these algebras are either computationally incomplete or support optimisations for only a subset of their operators. Also, some provide only limited facilities for optimising user defined data types; while others allow few algebraic ....

Lieuwin, D. and Dewitt, D. A transformation - based approach to optimizing loops in database programming languages, Proc. ACM SIGMOD, 1992.

....for or sets [INV91] were studied in [LW93b] and bag languages in [LW94c] see also [GM93] HKM93, HK94] show how to encode related database languages in the simply typed lambda calculus. A central issue is the optimization of collection type queries, see [Sch86, Osb88, Col89, HN91, LD91, LD92, Sar92] and [BK93, ACM93, CV93, CS93, Feg94, FMS93] Some connections between the s.r. monad and comprehension languages and equational transformations leading to optimizations are discussed in [BTBN91, BBW92, Won93] The important extension to object identifiers, which also allows recursive type ....

Daniel F. Lieuwen and David J. DeWitt. A transformation-based approach to optimizing loops in database programming languages. SIGMOD Record, 21(2):91--100, June 1992.

....performs all type checks before generating calls to the object manager interface. Release 1.1 of O , implementing the core functionality of the language, can be obtained from N. Gehani. Continuing work on O includes the investigation of performance related issues, 10 such as optimization [24], and the incorporation of new facilities into the language. These include support for large objects, versions [9] constraints and triggers [18] We are also studying the use of a persistent type catalog to store information about types of objects in an Ode Database. The design of such a catalog ....

D. G. Lieuwen and D. J. DeWitt, "A Transformation-Based Approach to Optimizing Loops in Database Programming Languages", Proc. ACM-SIGMOD 1992 Int'l Conf. on Management of Data, June 1992, 91-100.

....though these languages are beginning to disappear in favor to more declarative languages, such as OQL, there is a surge of interest to provide more computational power to existing declarative query languages without sacrificing performance. Consider for example the following O query (taken from [LD92] for (D of Divisions) for (E of Employees) suchthat (E division= D) f totpay = ChristmasBonus (E bpayD pftshare) 100; empcnt ; g These types of queries allow any kind of C code inside a for loop, including code that modifies the database. Earlier research on query optimization [LD92] ....

....[LD92] for (D of Divisions) for (E of Employees) suchthat (E division= D) f totpay = ChristmasBonus (E bpayD pftshare) 100; empcnt ; g These types of queries allow any kind of C code inside a for loop, including code that modifies the database. Earlier research on query optimization [LD92] has shown that queries in this form are very hard to optimize. Another problem to consider when sets and bags are combined with side effects is that the results may be unpredictable due to commutativity. For example, in the following O query for (e of Employees) x = e.salary the value of x ....

D. Lieuwen and D. DeWitt. A Transformation-Based approach to Optimizing Loops in Database Programming Languages. Proceedings of the ACM-SIGMOD International Conference on Management of Data, San Diego, California, pages 91--100, June 1992.

....well as disk les (to communicate temporary results) and then let the operating system do the scheduling has been identi ed as too costly an option [11, 12] The necessary process context switches, IPC cost, and I O of temporary results dominate the overall query cost. Dioeerent lines of research [7, 8, 16] led to the development of single process query engines in which the query plan is compiled into a single iterative or recursive procedure. Whenever possible these approaches strive for a stream based (or pipelined) execution to avoid I O of temporary data. The query processor bene ts from ....

....(unfolded) with the goal of fusing neighbouring combinators. Rule sets described if and how fusing was possible. The strategy then tried to fold the fused program back into combinator form, a step that involved complex pattern matching and, for some cases, could only be done semiautomatically. [16] developed a very speci c set of fusion rules for an imperative programming language in which plans had to be coded. To arrange a query plan as a sequence of combinators actually coincides very closely with the so called listful style of functional programming: complex list manipulations are ....

D.F. Lieuwen and D.J. DeWitt. A Transformation Based Approach to Optimizing Loops in Database Programming Languages. In Proc. of the ACM SIGMOD Int'l Conference on Management of Data, pages 91100, San Diego, California, June 1992.

....a large number of oids relative to n. For instance, for n=32, if each setvalued attribute contains 32 oids, a i r is expected to be 1.57 even if no clustering is attempted and references are randomly distributed across the n nodes. that employs our algorithms (and our program transformations [LIEU92a, LIEU92b]) 7. ....

Daniel Lieuwen and David DeWitt, A Transformation-based Approach to Optimizing Loops in Database Programming Languages. Proc. 1992 SIGMOD, June 1992.

....for processing the bulk data structures of OODBSs and DBPLs. We would like to simulate the algorithms described in this paper to validate our cost formulas. Finally, we would like to build a parallel version of a DBPL (probably O ) that employs our algorithms (and our program transformations [LIEU92a, LIEU92b]) 7. ....

Daniel Lieuwen and David DeWitt, A Transformation-based Approach to Optimizing Loops in Database Programming Languages. Proc. 1992 SIGMOD, June 1992.

....partition was used for the transaction log, and the other was used to store normal data. The client had a 2 megabyte (500 4K page) buffer pool; the server had a 600K (150 4K page) buffer pool. 11 The schema was as follows: ################################ 10 Earlier experiments reported in [LIEU92] were run on Version 1.2 of the EXODUS Storage Manager. We moved to the newer version to have access to the B trees provided by the new release. 11 In [LIEU92] we used a 10 megabyte buffer pool on a DECstation 3100, and our buffer pool could hold approximately 70 of the database. To make our ....

....a 600K (150 4K page) buffer pool. 11 The schema was as follows: ################################ 10 Earlier experiments reported in [LIEU92] were run on Version 1.2 of the EXODUS Storage Manager. We moved to the newer version to have access to the B trees provided by the new release. 11 In [LIEU92], we used a 10 megabyte buffer pool on a DECstation 3100, and our buffer pool could hold approximately 70 of the database. To make our examples more realistic, we wanted to decrease the fraction of the database that fits in the buffer pool. Rather than increase the size of the database, we made ....

[Article contains additional citation context not shown here]

Daniel Lieuwen and David DeWitt, A Transformation-based Approach to Optimizing Loops in Database Programming Languages. Proc. 1992 SIGMOD, June 1992.

No context found.

Lieuwin, D. and DeWitt, D. A transformation - based approach to optimizing loops in database programming languages, Proc. ACM SIGMOD, San Diego, June 1992, pp 91-100.

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