Cherniack, M., & Zdonik, S. B. (1996). Rule languages and internal algebras for rule-based optimizers. SIGMOD Record (ACM Special Interest Group on Management of Data), 25 (2), 401--412.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....we propose here a foundation for a systematic and quite general approach encompassing the old relational theory and a significant class of non relational queries, constraints and views. A novel property of our approach, one that we hope to exploit in relation to rule based optimization as in [CZ96], is that optimizing under constraints or deriving other constraints can be done within the equational theory of our internal framework by rewriting with the constraints themselves. In this paper, we use the terms constraints and dependencies interchangeably. In this opening section we plan to ....

M. Cherniack and S. B. Zdonik. Rule languages and internal algebras for rule-based optimizers. In Proceedings of the SIGMOD International Conference on Management of Data, pages 401--412, Montreal, Quebec, Canada, 1996.

....used here. Comparison with rule based optimizers. The C B strategy is extensible in the sense that one can add constraints to the logical and or physical schema and the optimizer need not be modified. This extensibility is in the same spirit with the extensibility of rule based systems [CZ96, GCD 94, HFLP89] However, in their case, one has to add rules, possibly with code (thus complex) to extend the capabilities of the optimizer. Extensibility in our case means just the addition of 188 new constraints and is thus at a higher level of abstraction (easier to use) Moreover, the ....

M. Cherniack and S. B. Zdonik. Rule languages and internal algebras for rule-based optimizers. In Proceedings of the SIGMOD International Conference on Management of Data, pages ??--??, Montreal, Quebec, Canada, 1996.

....[47] for information integration, where it turns out that the techniques presented in [15, 30, 39, 38] are neither general enough nor flexible enough to be adapted to the problems we wish to solve. Moreover, we present our technique in a form that is easy to integrate in the rule based paradigm [17], and easy to combine with conventional optimization techniques [41] such as selection pushing and join reordering. Theoretical aspects We prove that our method is complete, i.e. finds the query plans that are minimal in a precise sense, for path conjunctive(PC ) queries and physical access ....

....that P 0 , P 1 , P 2 , and P 3 are obtained as minimal queries in this algorithm. Steps (3) and (4) choose the cheapest plan among them. Rule based implementation In an implementation, the conceptual search of algorithm 3. 1 can be specified implicitly by configuring a rule based optimizer ([17, 23]) with the two rewrite rules chase and backchase, and requesting that the application of the chase rule always takes precedence over that of the backchase rule. Depending on the search strategy implemented by the optimizer, the search space may not be explored exhaustively but rather pruned using ....

M. Cherniack and S. B. Zdonik. Rule languages and internal algebras for rule-based optimizers. In Proceedings of the SIGMOD International Conference on Management of Data, pages ??--??, Montreal, Quebec, Canada, 1996.

....together in a query tree and then the Remote Join Eval rule would collapse the explicit join operator and the two retrieves into a single retrieval of a remote join. More complex examples of this sort of rewriting rule programs are presented in the query optimizer for object oriented languages in [11,12]. Second, another area for further research is the interplay of query planning and plan execution. There are several types of interplay that have been explored in the database literature, but we believe that a general planning framework such as PbR will provide a more principled and general ....

M. Cherniack, S.B. Zdonik, Rule languages and internal algebras for rule-based optimizers, SIGMOD Record (ACM Special Interest Group on Management of Data) 25 (2) (1996) 401--412.

....while applying different optimization strategies like materialization or cost based evaluation of semantically equivalent filter graphs. Rule based transformation and optimization of operator graphs have been studied for more than a decade in the context of extendible database query optimizers [RH87, CZ96], optimizer generators [SS90, GM93] and query optimizers for object oriented databases [VD91] Structurally, filter graphs are nearly identical to operator graphs. Semantically, of course, they are different there exist, however, several analogies: Logic operators (e.g. join) correspond to ....

Cherniack, M., Zdonik, S. B.: Rule Languages and Internal Algebras for Rule-Based Optimizers. In: Proc. of the 1996 ACM SIGMOD Int. Conf. on Management of Data (Montreal, Canada, June 4--6), SIGMOD Record Vol. 25, Issue 2, June 1996, pp. 401-- 412.

....together in a query tree and then the Remote Join Eval rule would collapse the explicit join operator and the two retrieves into a single retrieval of a remote join. More complex examples of this sort of rewriting rule programs are presented in the query optimizer for object oriented languages in [11, 12]. Second, another area for further research is the interplay of query planning and plan execution. There are several types of interplay that have been explored in the database literature, but we believe that a general planning framework such as PbR will provide a more principled and general ....

M. Cherniack and S. B. Zdonik. Rule languages and internal algebras for rule-based optimizers. SIGMOD Record (ACM Special Interest Group on 49 Management of Data), 25(2):401--412, June 1996.

....discounted 3 queries. Performance analysis of the technique, and experimental results over a TPC D benchmark database in DB2, are presented in Section 4. Section 5 concludes with issues and future work. 1. 2 Related Work There is a substantial body of research in rewrite based query optimization [4, 6, 7, 19]. However, all of the techniques discussed in literature considered rewriting a query into a logically equivalent form. This is a different goal than the one we consider here: we are interested in generating and efficiently evaluating a query (view) from which an unfolding has been removed. Thus, ....

M. Cherniack and S. Zdonik. Rule languages and internal algebras for rule based optimizers. In Proc. SIGMOD, pages 401--412, 1996.

....rule selection. Exceptions are Volcano [McK93, GM93] EXODUS [GD87] and Gral [BG92] cf. Section 3) The rule selection becomes more interesting if more rules are present in the rule base. Newer languages define more operators resulting in more equivalences, hence more rules (e.g. KOLA [Che96, CZ96, CZ98] consists of about 60 operators and all in all about 660 rules) In the worst case, all of these rules would have to be checked for matching by the term rewriting system in each rewriting step. This worst case may be prevented, e.g. by hashing or classifying the rules in an appropriate ....

....the system is not as efficient as necessary. This may be due to the lack of any heuristical decision. On the other hand, EXODUS is flexible due to the use of statistics, and extending the rule base does not result in the need of much programming. COKO KOLA is the combination of the algebra KOLA [CZ96, Che96] with the language COKO [CZ98] for transformation control. COKO provides a framework for the programming of rule applications and query tree traversals, thus the rule ordering has to be programmed. In this way, the rule ordering is fixed according to COKO language constructs. Since the ....

M. Cherniack and S.B. Zdonik. Rule Languages and Internal Algebras for Rule-Based Optimizers. In Proc. of the ACM SIGMOD Int. Conference on Management of Data, Montreal, Quebec, June 1996.

....with optimizing rewrite systems. A specific translation technique for the decomposed model can be found in [28,45] For supporting object oriented systems, database researchers have tried to repeat the successes in the relational field by proposing a number of object oriented query algebras [41,8]. They offer a nested object data model for supporting complex objects and support multiple collection types like Set, List and Bag. These languages were designed as input languages for algebraic OQL query SELECT year, sum(total) FROM ( SELECT price tax AS total, year(item.order.day) AS year ....

M. Cherniack and S.B. Zdonik. Rule languages and internal algebras for rule-based optimizers. In Proc. SIGMOD Conf., pages 401--412, Atlantic City (NJ), June 1996.

....a trend towards semantically simpler operators since their optimization, i.e. the formulation of equivalences for these operators seems to be more promising and much simpler (and therefore also meant to be more efficient) than the optimization of complex operators. In this way, algebras like KOLA [CZ96, Che96] are developed, consisting of about 60 operators with nearly 660 equivalence rules. The naive approach of applying (or even matching) all equivalence rules against all operator nodes in a given query tree leads to a combinational explosion. Also, the logical search space of equivalent ....

....Moreover, in [Fre87] the search space is spanned exhaustively; this is neither efficient nor ever really feasible in an appropriate amount of time. Our approach of the heuristic search space generation will remedy things here, too. COKO KOLA is a combination of the already mentioned algebra KOLA [CZ96, Che96] with the language COKO [CZ98] for transformation control. COKO provides a framework for the programming of rule applications and query tree traversals. Unfortunately, the application order of rules has to be fixed by programming code and therefore, the system loses some of its possible ....

M. Cherniack and S.B. Zdonik. Rule Languages and Internal Algebras for Rule-Based Optimizers. In Proc. of the ACM SIGMOD Int. Conference on Management of Data, Montreal, Quebec, June 1996.

....primitives suffices for implementing the quasi totality of ODMG ODL OQL [Cat96] Using boolean aggregates, CoDi can represent dependencies as equalities between boolean valued queries. An important property of our approach, one that we hope to exploit in relation to rule based optimization as in [CZ96] is that optimizing under dependencies or deriving other dependencies can be done within the equational theory by rewriting with the dependencies themselves. Overview The types, expressions, and basic equational laws of CoDi are introduced in section 2. Appendix A contains the rest of the ....

M. Cherniack and S. B. Zdonik. Rule languages and internal algebras for rule-based optimizers. In Proceedings of the SIGMOD International Conference on Management of Data, pages ??--??, Montreal, Quebec, Canada, 1996.

.... Function Semantics to Optimize Queries Mitch Cherniack Brown University Providence, RI 02912 mfc cs.brown.edu Stan Zdonik Brown University Providence, RI 02912 sbz cs.brown.edu Abstract The goal of the COKO KOLA project [10, 9] is to express rules of rule based optimizers in a manner permitting verification with a theorem prover. In [10] we considered query transformations that were too general to be expressed with rewrite rules. In this paper, we consider the complementary issue of expressing query transformations ....

....Mitch Cherniack Brown University Providence, RI 02912 mfc cs.brown.edu Stan Zdonik Brown University Providence, RI 02912 sbz cs.brown.edu Abstract The goal of the COKO KOLA project [10, 9] is to express rules of rule based optimizers in a manner permitting verification with a theorem prover. In [10], we considered query transformations that were too general to be expressed with rewrite rules. In this paper, we consider the complementary issue of expressing query transformations that are too specifc for rewrite rules. Such transformations require rewrite rules to be supplemented with ....

[Article contains additional citation context not shown here]

M. Cherniack and S. B. Zdonik. Rule languages and internal algebras for rule-based optimizers. In Proc. ACM SIGMOD Int'l Conference on Management of Data, Montreal, Quebec, Canada, June 1996.

....because an optimizer s behavior can be modified by changing its rule set. But this is difficult unless rules are straightforward to understand and reason about. Therefore, extensibility is best achieved by expressing rules declaratively, rather than with code. e.g. with rewrite rules) In [5], we showed that the choice of query representation determines the effectiveness with which rewrite rules can express query rewrites. Rewrite rules use pairs of patterns to specify rewrites that identify and rearrange subexpressions. Variable based query representations (i.e. representations ....

....algorithms achieves a balance of understandability, expressivity and performance. A COKO transformation is understandable; it can be decomposed into its rewrite rules and even shown to preserve the semantics of the queries it transforms by proving the same property of these rules. As we showed in [5], this can be done with help from a theorem prover [10] At the same time, COKO firing algorithms make transformations capable of expressing a wide variety of efficient query rewrites, as we later show. This work generalizes and extends our KOLA work. COKO transformations behave like rewrite ....

[Article contains additional citation context not shown here]

M. Cherniack and S. B. Zdonik. Rule languages and internal algebras for rulebased optimizers. In Proc. ACM SIGMOD Int'l Conference on Management of Data, Montreal, Quebec, Canada, June 1996.

.... Function Semantics to Optimize Queries Mitch Cherniack Brown University Providence, RI 02912 mfc cs.brown.edu Stan Zdonik Brown University Providence, RI 02912 sbz cs.brown.edu Abstract The goal of the COKO KOLA project [10, 9] is to express rules of rule based optimizers in a manner permitting verification with a theorem prover. In [10] we considered query transformations that were too general to be expressed with rewrite rules. In this paper, we consider the complementary issue of expressing query transformations ....

....Mitch Cherniack Brown University Providence, RI 02912 mfc cs.brown.edu Stan Zdonik Brown University Providence, RI 02912 sbz cs.brown.edu Abstract The goal of the COKO KOLA project [10, 9] is to express rules of rule based optimizers in a manner permitting verification with a theorem prover. In [10], we considered query transformations that were too general to be expressed with rewrite rules. In this paper, we consider the complementary issue of expressing query transformations that are too specifc for rewrite rules. Such transformations require rewrite rules to be supplemented with ....

[Article contains additional citation context not shown here]

M. Cherniack and S. B. Zdonik. Rule languages and internal algebras for rule-based optimizers. In Proc. ACM SIGMOD Int'l Conference on Management of Data, Montreal, Quebec, Canada, June 1996.

No context found.

Cherniack, M., & Zdonik, S. B. (1996). Rule languages and internal algebras for rule-based optimizers. SIGMOD Record (ACM Special Interest Group on Management of Data), 25 (2), 401--412.

No context found.

M. Cherniack and S. B. Zdonik, "Rule languages and internal algebras for rule-based optimizers," Proceedings of the ACM SIGMOD International Conference on Management of Data. Montreal, Canada, June, 1996, pp.

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