CHOENNI, S ., BERTINO, E ., BLANKEN, H . M ., AND CHANG, T. On the Selection of Optimal Index Configuration in OO Databases. Proc. of the 10 ICDE, 1994.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....Christophides et al. 8] have studied this problem, their work focuses on query rewriting and not indexes, and our mechanism could utilize their techniques (or those of [15] to better optimize generalized path expressions over raw paths. Finally, OO indexes must deal with class inheritance [7], while XML indexes do not. Text indexing has been studied extensively in both structured and unstructured databases. Suffix arrays and compressed suffix arrays [16] based on Patricia tries, provide partial match searching rather than path navigation. Several data and query models for structured ....

S. Choenni et al. On the selection of optimal index configuration in OO databases. In Proc. ICDE, 1994.

....updated and thereby invalidated the stored result. The goal of keeping validity values is to facilitate lazy rematerialization of the function result. Indexing Techniques. We utilized the work by Bertino et al. regarding the performance of basic operations with traditional index organizations [2, 4]. 8 Conclusions To the best of our knowledge, ours is the first work to address the specific needs of the path query view problem for OODBs and to present a solution that is tailored to these needs. We introduce a new Satisfiability Indicating Multi Index (SMX) organization, which maintains ....

S. Choenni, E. Bertino, H. M. Blanken, and T. Chang. On the selection of optimal index configurations in OO databases. In IEEE Int. Conf. on Data Engineering, pages 526--537, 1994.

....Techniques. We also considered the work of previous researchers who have compared index structures when designing our SMX organization. In particular, the work of Bertino et al. informed us of the costs and issues involved in the performance of basic operations with traditional index organizations [2, 3]. 6 Conclusions and Future Work To the best of our knowledge, ours is the first work to identify and address the specific needs of the path query view problem for object oriented databases and to present a solution that is tailored to these needs. We introduce a new Satisfiability Indicating ....

S. Choenni, E. Bertino, H. M. Blanken, and T. Chang. On the selection of optimal index configurations in OO databases. In IEEE International Conference on Data Engineering, pages 526--537, 1994.

....updated and thereby invalidated the stored result. The goal of keeping validity values is to facilitate lazy rematerialization of the function result. Indexing Techniques. We utilized the work by Bertino et al. regarding the performance of basic operations with traditional index organizations [2, 4]. 8 Conclusions To the best of our knowledge, ours is the first work to address the specific needs of the path query view problem for OODBs and to present a solution that is tailored to these needs. We introduce a new Satisfiability Indicating Multi Index (SMX) organization, which maintains ....

S. Choenni, E. Bertino, H. M. Blanken, and T. Chang. On the selection of optimal index configurations in OO databases. In IEEE Int. Conf. on Data Engineering, pages 526--537, 1994.

....So an important aspect of the optimization of path queries is the selection of indices on the path. A large database schema may include several hundred classes and attributes, so that the set of index structures defined for the corresponding database can be very large. Different authors [6, 9, 10] discuss how to configure profitable index structures for a given large database schema. With this background, determining the optimal query execution strategy itself can become very expensive for two reasons. Many alternative indices may be available to evaluate the same path and the applicable ....

CHOENNI, S., BERTINO, E., BLANKEN, H. M., AND CHANG, T. On the Selection of Optimal Index Configuration in OO Databases. Proc. of the 10 th ICDE, 1994.

....will discuss extensions to our SMX solution that allow us to relax these assumptions in Section 7. We calculate cost in terms of the number of page accesses, and assume that a page contains objects of only one class. Table 2 presents the system parameters, adapted from the work of Choenni et al. [4] and Korth and Silberschatz [9] used in the following equations. Table 3 lists the path query view specific parameters used in our equations. parameter definition CRIRX Cost of retrieving an index record with organization X . CMIRX Cost of maintaining an index record with organization X. CP ....

....or maintaining a single index record in an SMX organization is thus very similar to that of performing the corresponding operation under an MX organization. As other researchers have previously discussed and contrasted the cost of performing basic functions using traditional index organizations [2, 4], we confine our model of the costs of performing basic operations to a comparison between the SMX and MX operations. 5.1.1 Storage Costs Each SMX index record extends the MX index record with additional information as described in Section 4. Each leaf node used to store an index record from the ....

[Article contains additional citation context not shown here]

S. Choenni, E. Bertino, H. M. Blanken, and T. Chang. On the selection of optimal index configurations in OO databases. In IEEE International Conference on Data Engineering, pages 526--537, 1994.

....records. An auxiliary index, which basically keeps direct reference information between objects, together with the extra information in the primary records are used to propagate updates. The nested index structure in general outperforms the other two index structures [3, 2] Choenni et al. [7] propose an optimal index configuration by splitting a long path expression into shorter ones, and by indexing the shorter paths with existing index structures such as those in [4, 2] Shekita and Carey [30] describe a mechanism called field replications which replicate the values of the nested ....

S. Choenni, E. Bertino, H. M. Blanken, and T. Chang. On the selection of optimal index configuration in OO databases. In Proc. Int. Conf. Data Engineering, pages 526--537, Phoenix, AZ, USA, February 1994.

....separate structures to redundantly store objects which are frequently traversed by database queries. A hybrid indexing technique, called a generalized index, is proposed in [8] to support class hierarchy with complex and primitive objects. Indexing in GemStone OODB is described in [16] In [6], an optimal index configuration for a path is achieved by splitting the path into subpaths and optimally indexing each subpath. It is noted that most of these prior works only considered indexing along a single path in a nested object hierarchy. However, as will be shown later, the effects of two ....

S. Choenni, E. Bertino, H. M. Blanken, and T. Chang. On The Selection of Optimal Index Configuration in OO Databases. Proceedings of the 10th International Conference on Data Engineering, pages 526--537, February 1994.

....String String Transmission String String String String String String Numeric VehicleBody Numeric Numeric Numeric VehicleDrivertrain Company Vehicle Division PistonEngine Figure 1: An example for nested object hierarchy. and primitive objects. Indexing in GemStone OODB is described in [13] In [6], an optimal index configuration for a path is achieved by splitting the path into subpaths and optimally indexing each subpath. It is noted that most of these prior works only considered indexing along a single path in a nested object hierarchy. However, as will be shown later, the effects of two ....

S. Choenni, E. Bertino, H. M. Blanken, and T. Chang. On The Selection of Optimal Index Configuration in OO Databases. Proceedings of the 10th International Conference on Data Engineering, pages 526--537, February 1994.

....is very complex, the path dictionary can be decomposed into a number of simpler path dictionaries. For instance, a long path may be decomposed into several small path segments for design and efficiency reasons. The configuration issues involved with path and nested indexes were discussed in [6]. In this paper, we assume that only one path dictionary is built for an aggregation hierarchy. 3.2 Attribute Index While the path dictionary supports fast traversal among objects, it by itself will not help predicate evaluation which involves finding objects meeting certain conditions specified ....

....belong to C n , is the terminal object of the 2 Although i denotes the OID of an object, we use it to refer to the object itself, as in this case, when no confusion arises. Path = Person.Vehicle.Company Company[1] Vehicle[5] Person[3] Person[7] Vehicle[12] Person[4] Company[2] Vehicle[6]( Vehicle[9] Vehicle[11] Company[3] Vehicle[3] Company[4] Vehicle[4] Vehicle[7] Person[1] Person[6] Company[5] Vehicle[1] Person[2] Vehicle[2] Person[8] Person[12] Vehicle[8] Person[5] Vehicle[10] Person[9] Figure 5: Examples of the s expression scheme. paths denoted ....

[Article contains additional citation context not shown here]

S. Choenni, E. Bertino, H.M. Blanken & T. Chang, "On the Selection of Optimal Index Configuration in OO Databases," Proceedings of the Tenth International Conference on Data Engineering , Houston, TX, 1994, 526--537.

....classes are large and for IIX and MIX if the number of classes involved in the inheritance hierarchy and the cardinality of these classes are large. It should be clear that if the whole index record has to be retrieved pr X = dln X =pe. A procedure to compute the height of an index can be found in [7] The maintenance cost of a single index record can be expressed by: CML(hX ; pmX ) ae hX 1 ln X p hX Gamma 1 2 pmX otherwise in which pmX is the average number of pages involved in maintaining an index record with regard to an index organization X . The extra page in the above formula ....

....The value pmX may differ depending on the fact whether an index should be maintained due to a deletion or an insertion. Whenever this is the case we use the notation pmdX and pmiX respectively to distinguish these situations. For more details with regard to these parameters, we refer to [7]. The second pair of functions concerns the maintenance and retrieval cost of a set of index records, denoted as CRT and CMT respectively. To retrieve t k index records at level k of an index from n k records stored on p k pages the number of page accesses (npa) can be estimated by the formula of ....

[Article contains additional citation context not shown here]

Choenni, R., Bertino, E., Blanken, H.M., Chang, S.C., On the Selection of Optimal Index Configuration in OO Databases (extended version of this paper), to appear as tecnical report.

No context found.

CHOENNI, S ., BERTINO, E ., BLANKEN, H . M ., AND CHANG, T. On the Selection of Optimal Index Configuration in OO Databases. Proc. of the 10 ICDE, 1994.

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