P. Aoki. Implementation of Extended Indexes in POSTGRES. SIGIR Forum, 2(1):2--9, 1991.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....implementation. As we will see in Chapter 3, the level of abstraction offered by the interface 1 As with most other object relational mechanisms, the AM extensibility mechanisms present in those commercial systems are direct descendents of those pioneered in research systems such as Postgres [Aok91] 2 See [BSSJ99] for an account of an implementation of an AM using IDS UDO s VII. 9 Storage, Buffer, Log, Query Processing B tree New AMs require custom CC R code Query Processing Storage, Buffer, Log, SR treeSR tree R tree R tree (a) Standard ORDBMS (b) ORDBMS with ....

....function entry points of the adapter need to be communicated to the AM extension. In a built in, datatypeextensible AM such as the server provided R tree, this is achieved through an operator class, or opclass, a named collection of UDFs that essentially contains the type adapter s functions (see [Aok91] for details) In this architecture, the external, datatype specific functions that the built in AM needs to call are resolved using the server supplied mechanisms for registered UDFs and called as UDFs. In contrast, the GiST based architecture avoids these UDF calls and uses standard C function ....

P. Aoki. Implementation of Extended Indexes in POSTGRES. SIGIR Forum, 2(1):2--9, 1991.

....to the original version, which shows that it is not necessary to use custom software to get good performance out of an IR system. 41 Conclusions An increasing tendency to use off the shelf storage management software for the implementation of IR systems can be noted in the literature (e.g. [49, 8, 7, 15, 1, 16, 22, 17]) Although custom storage management software provides high performance, its inherent inflexibility and high development costs are no longer considered acceptable. However, it is not yet clear what kind of storage management software should be used instead of custom solutions. For IRF 1 two ....

Aoki, Paul M. 1991. "Implementation of Extended Indexes in POSTGRES. " SIGIR Forum. 25(1): 2--9.

....concerned the access method design and was the decision to support indexing only on the value of a field and not on a function of a value. The utility of indexes on functions of values is discussed in [LYNC88] and the capability was retrofitted, rather inelegantly, into one version of POSTGRES [AOKI89]. Another comment on the access method design concerns extendibility. Because a user can add new base types dynamically, it is essential that he also be able to add new access methods to POSTGRES if the system does not come with an access method that supports efficient access to his types. The ....

Aoki, P., "Implementation of Extended Indexes in POSTGRES," Electronics Research Laboratory, University of California, Technical Report 89-62, July 1989.

....k D B trees [Rob81] and hB trees [LS90] Multidimensional data can also be transformed into unidimensional data using a spacefilling curve [Jag90] after transformation, a B tree can be used to index the resulting unidimensional data. Extensible key indices were introduced in POSTGRES [Sto86, Aok91] and are included in Illustra [Ill94] both of which have distinct extensible B tree and R tree implementations. These extensible indices allow many types of data to be indexed, but only support a fixed set of query predicates. For example, POSTGRES B trees support the usual ordering ....

P. M. Aoki. Implementation of Extended Indexes in POSTGRES. SIGIR Forum, 25(1):2--9, 1991.

....k D B trees [Rob81] and hB trees [LS90] Multidimensional data can also be transformed into unidimensional data using a space filling curve [Jag90] after transformation, a B tree can be used to index the resulting unidimensional data. Extensible key indices were introduced in POSTGRES [Sto86, Aok91] and are included in Illustra [Sto93] both of which have distinct extensible B tree and R tree implementations. These extensible indices allow many types of data to be indexed, but only support a fixed set of query predicates. For example, POSTGRES B trees support the usual ordering ....

P. M. Aoki. Implementation of Extended Indexes in POSTGRES. SIGIR Forum, 25(1):2--9, 1991.

....computed value of a column. The new List datatype is used to accomplish this extension. Their proposal provides much of the foundation to the implementation discussed in this paper. Aoki implemented an approach to extended indexing along the ideas of Lynch and Stonebraker with only slight changes [1]. Some of Aoki s work was restored from an earlier version of POSTGRES, though most was modified for this implementation. Aoki created an fbtree (functional B tree) access method, similar to a regular B tree, that supports multiple index keys from one column value. This code was recovered and ....

....retrieval effectiveness. 4.2 Index Keywords A new access method, the fbtree, was created to handle the indexing. The fbtree is very similar to the regular btree access method with the insert and the search routines customized to support the new indexing. The fbtree was first developed by Aoki [1] and modified here to incorporate probabilistic retrieval. The insert routine breaks up the document into indexable terms terms that have successfully passed the tokenize routine, and inserts one index record for each unique term in a document. The index records consists of the index term and ....

Aoki, P. "Implementation of Extended Indexes in POSTGRES," SIGIR Forum, Spring 1991, vol.25, (no.1):2-9.

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