K. Nrvag. The Persistent Cache: Improving OID indexing in temporal objectoriented database systems. In Proceedings of the 25th VLDB Conference, 1999.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....necessary, and the additional update cost is only marginal. However, update and insert into such databases can in general be quite expensive. We believe that further work should focus on reducing these costs, for example using an approach based on structures like LHAM [16] and the Persistent Cache [17] access methods. 45 50 55 60 65 70 75 80 85 Size of version database MB No TimeIdx TVArray VPArray (a) Update set of web pages. 75 80 85 90 95 100 105 110 115 Size of version database MB No TimeIdx TVArray VPArray (b) Insert new set of web pages. 1.8 2 2.2 2.4 2.6 ....

K. Nrvag. The Persistent Cache: Improving OID indexing in temporal object-oriented database systems. In Proceedings of the 25th VLDB Conference, 1999.

....be used together with simulations, to determine which parameters should be used in the simulations, and to help explain the results of the simulations. Our application of the models presented in this paper has in particular been OID indexes in object database systems, where we have successfully [8,9,11,12] employed the buffer models in our analysis. We have also deeply appreciated analytical modeling as a powerful tool in the design of Vagabond [10] Several bottlenecks were identified using the analytical models, and many of the solutions were also developed with the help of the models. Another ....

K. Nrvag. The Persistent Cache: Improving OID indexing in temporal objectoriented database systems. In Proceedings of the 25th VLDB Conference, 1999.

....4. Overview of the TreeIDX, PCache, and index related main memory buffers. PCache nodes PC1, PC4 and PC6, and three TreeIDX nodes (denoted TNODn) are in the main memory buffer. larly checkpointed to disk. A more detailed description and performance analysis of the PCache can be found in [12]. In order to avoid confusion, we will in the rest of this paper denote the OID index tree itself as the TreeIDX, and use OIDX to denote the combined index system, i.e. the PCache and the TreeIDX. Thus, when we say an entry is in the OIDX, it can be in the PCache, in the TreeIDX, or in both. ....

K. Nrvag. The Persistent Cache: Improving OID indexing in temporal object-oriented database systems. In Proceedings of the 25th VLDB Conference, 1999.

....to the text indexing cost than in the document database case. Update and insert into document databases are in general quite expensive. We believe that further work should focus on reducing these costs, for example using an approach based on structures like the LHAM [16] or the Persistent Cache [17]. Acknowledgments Many of the basic ideas of this paper was developed during the 9 months the author spent as an ERCIM fellow in the Verso group at INRIA, France, in 2001. The main part was done when the author visited Athens University of Economics and Business in 2002, supported by grant ....

K. Nrvag. The Persistent Cache: Improving OID indexing in temporal object-oriented database systems. In Proceedings of the 25th VLDB Conference, 1999. 20

....because either 1) the previous or 2) the new version must be written to a new place. Thus, when supporting temporal data management, the indexing cost is the same in these two approaches. We have in previous work developed several techniques that can be used to reduce the OIDX access cost [9, 14]. 4 3.2 Objects in Vagabond In our approach, all objects smaller than a certain threshold are written as one contiguous object. Objects larger than this threshold are segmented into subobjects, and a subobject indexis maintained for each of these large objects (this is done transparently for ....

....bottleneck, and in this section, we present the persistent cache (PCache) an approach that reduces the index update and lookup costs in temporal ODBs. To make this paper self containing, we now describe briefly the PCache. A more detailed description and performance analysis can be found in [9]. 3.4 The PCache The ODs in an OIDX have in general a low locality. The OD cache makes read accesses efficient, but in a database with many objects, most of the ODs that are updated during one checkpoint interval 2 will reside in different leaf nodes. This low locality means that many leaf ....

[Article contains additional citation context not shown here]

K. Nrvag. The Persistent Cache: Improving OID indexing in temporal object-oriented database systems. In Proceedings of the 25th VLDB Conference, 1999.

.... To reduce the indexing cost, we have developed a new OID index structure for parallel temporal OODBs [7] as well as several novel techniques that reduce the cost of OID indexing: 1 ffl Writable object descriptor cache for temporal OODBs [10] ffl Persistent caching of OID index entries [6]. A second bottleneck is object retrieval. The Vagabond system is write optimized, and as a result, object retrieval and index lookup can become a serious bottleneck. We will use some techniques to reduce the number and size of read operations, which can improve object retrieval performance ....

....compression. Compression will also improve write efficiency, as it reduces the amount of data needed to be written to disk. We will now give an overview over the techniques. For a more detailed description, we refer to the corresponding papers where these techniques are presented and analyzed [5, 6, 10]. 4.1 Writable Object Descriptor Cache To reduce disk I O, the most recently used index pages are kept in an index page buffer. OIDX pages will in general have low locality, and to increase the probability of finding a certain OD needed for a mapping from OID to physical address, it is also ....

K. Nrvag. The Persistent Cache: Improving OID indexing in temporal object-oriented database systems. In Proceedings of the 25th VLDB Conference, 1999.

....a temporal ODB on top of O 2 . In a previous paper, we have studied OIDX lookup cost and optimal use of memory in a nontemporal ODB [7] and OIDX lookup and update cost in the context of a temporal ODB in [8] We have also shown how the OIDX costs can be reduced by using a persistent cache [5], an intermediate storage structure on disk. 3 An Overview of Vagabond In this section we provide a brief overview of the most important aspects of Vagabond in order to set the work reported in this paper into context. For more comprehensive overview and references, we refer to [6] 3.0.1 ....

K. Nrvag. The Persistent Cache: Improving OID indexing in temporal object-oriented database systems. In Proceedings of the 25th VLDB Conference, 1999.

No context found.

K. Nrvag. The Persistent Cache: Improving OID indexing in temporal objectoriented database systems. In Proceedings of the 25th VLDB Conference, 1999.

No context found.

K. Nrvag. The Persistent Cache: Improving OID indexing in temporal objectoriented database systems. In Proceedings of the 25th VLDB Conference, 1999.

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