C. D. French. Teaching an OLTP Database Kernel Advanced Data Warehousing Techniques. In IEEE ICDE, pages 194--198, 1997.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....as an additional rule in the query rewriting phase. The observations we summarized in the experimental section (Section 7) can be used to guide the design of those heuristic rules. We aim to use the adaptation technique in systems that are queried more frequently than updated. As argued in [Fre95, Fre97] an OLAP system separated from the traditional OLTP system can provide much better query performance. In such systems, updates are batched and data are changed only periodically. As a result, we assume that there are no data changes during view adaptation in this article. When batch updates ....

Clark D. French. Teaching an OLTP database kernel advanced data warehousing techniques. In Proc. IEEE Int'l Conf. on Data Eng., pages 194--198, 1997.

....of indexing algorithms given in [LC86b] would be valid on today s architectures. In fact, our experimental results indicate very di#erent relative outcomes from [LC86b] for lookup speed. Another recent development has been the explosion of interest in On Line Analytical Processing (OLAP) Fre95, Fre97] contrasts the requirements of OLAP with OLTP systems and contends that the real performance gains can be obtained by separating the two systems. A dedicated OLAP system can have a much better query performance if we are willing to sacrifice update performance. Commercial systems such as Sybase ....

Clark D. French. Teaching an OLTP database kernel advanced data warehousing techniques. In Proc. IEEE Int'l Conf. on Data Eng., 1997.

....way to use a large amount of RAM. Instead, a specialized main memory storage unit should be built to take advantage of the fact that all the data reside persistently in RAM. LSC92] reported a fourfold speed improvement when using a specialized main memory resident storage component. Fre95, Fre97] contrasted the requirements of OLAP with OLTP systems and contended that the real performance gains can be obtained by separating the two systems. A dedicated OLAP system can have a much better query performance if we are willing to sacrifice the update performance. For example, if only batch ....

Clark D. French. Teaching an OLTP database kernel advanced data warehousing techniques. In Proc. IEEE Int'l Conf. on Data Eng., pages 194--198, 1997.

....could also potentially be applied to conventional RDWMSs, storage size reductions are likely to be much higher with Curio. Indeed, most compression algorithms will yield higher compression ratios if the set of values being compressed ranges over a small domain (and thus displays repetitions) [11]. This is more likely to happen with a vertical partitioning approach than with a conventional table, where the data is stored row by row. Moreover, while it will sometimes be necessary to decompress the data, there exist query processing algorithms that can operate on compressed data. We can ....

C.D. French. Teaching an OLTP database kernel advanced datawarehousing techniques. In Proc. 13th ICDE, pages 194--198, Birmingham, UK, April 7-11 1997.

....column in the fact table results in a duplication of data. In such situations, it is advisable to only store the index if original table records can be reconstructed easily from the index itself. This is the starting point of the proposed DataIndex scheme and is how Sybase IQ stores data [19, 46]. Furthermore, with DataIndexes, each BDI of a table is stored separately with ordinal positon based mapping providing more efficient access to individual record fields compared to other vertical partitioning based methods. Because BDIs are stored separately, only columns of interest need to be ....

....however, are likely to be negligible when compared with the overhead incurred by expensive index update operations. Compressibility: It should be clear that DataIndexes can be compressed much more readily than conventional tables, since the range of values each DataIndex covers is much smaller [19]. Some compression techniques allow operations on the underlying data to be performed without decompression. Thus, we can expect that compressed DataIndexes can provide even better performance than those studied in this paper. Other OLAP queries: The results in [46] lead us to believe that ....

C.D. French. Teaching an OLTP database kernel advanced datawarehousing techniques. In Proc. 13th ICDE, pages 194--198, Birmingham, UK, April 7-11 1997.

....of the fact table, storing both the index and the corresponding column in the fact table results in a duplication of data. In such situations, it is advisable to only store the index if original table records can be reconstructed easily from the index itself. This is how Sybase IQ stores data [8, 16]. In what follows, we extend the original notion of the projection index to allow a single projection index to contain multiple columns. A graphical representation of this structure is shown in Figure 2. In this figure, we show the actual storage configurations of the two cases: a base table ....

C.D. French. Teaching an OLTP database kernel advanced datawarehousing techniques. In Proc. 13th ICDE, pages 194--198, Birmingham, UK, April 7-11 1997.

No context found.

C. D. French. Teaching an OLTP Database Kernel Advanced Data Warehousing Techniques. In IEEE ICDE, pages 194--198, 1997.

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