Abstract: Server-integrated methods and apparatus for processing precomputed views for answering user queries on a database are described. In accordance with one embodiment, a system for answering user queries on a database stored on a server includes a query processing system integrated into the server and configured to respond to user queries, and a query processing system integrated into the server and configured to process precomputed results of user queries on the database. In another aspect, a method of answering user queries on a database stored on a server includes responding to user queries received at the server, and processing, at the server, precomputed results of user queries on the database.

Abstract: A data replication processing system includes a source database system and one or more target database systems all of which are coupled through a network. The replication processing system provides a replication protocol that allows for the source database and the target databases to remain consistent and in agreement. A method for the replication processing system includes having a target database request a refresh of its subset of the source database from the source database system. The source database system receives the refresh request message and constructs a refresh reply message that includes transactions from the source and the target. The target database system receives the refresh reply message from the source database system and applies the changes to its subset of the source database so that the two databases are in agreement with respect to their appropriate data sets.

Abstract: A computer operated apparatus for generating a visual information system is disclosed. A virtual world associated with an application is built using building blocks such as scenes, data sources, global parameters, and resources. A scene is a visual display of information much like a presentation slide, except that the information may be linked to data stored in a database or other data storage systems. Within a scene, values resulting from a data source are represented graphically as user-defined data elements. Data sources are built with a block diagraming tool which generates one or more database queries. The queries may be SQL queries. Scenes are created with a drawing editor which transparently binds data sources to the graphical elements of the scenes. When the virtual world is completed, an execution image of the virtual world may be represented as byte code. The byte code representing the virtual world may be executed by a runtime control to provide desired information to users.

Abstract: Methods and apparatus for processing precomputed views for answering user queries on a database are described. In accordance with one aspect, a user database query on precomputation strategy effectiveness is responded to by defining, based upon user database query history, an analysis space consisting of a subset of all possible views for the database, and by characterizing the views in the analysis space. In accordance with a second aspect, a structure is imposed on an analysis space consisting of a subset of all possible views for the database based upon the capabilities of a query rewriting facility, and the views in the analysis space are characterized. In accordance with a third aspect, an analysis space consisting of a subset of all possible views for the database is defined, and a cost formula is applied to the analysis space based upon a user-defined subset of the data contained in the database.

Abstract: A computer operated apparatus for generating a visual information system is disclosed. A virtual world associated with an application is built using building blocks such as scenes, data sources, global parameters, and resources. A scene is a visual display of information much like a presentation slide, except that the information may be linked to data stored in a database or other data storage systems. Within a scene, values resulting from a data source are represented graphically as user-defined data elements. Data sources are built with a block diagraming tool which generates one or more database queries. The queries may be SQL queries. Scenes are created with a drawing editor which transparently binds data sources to the graphical elements of the scenes. When the virtual world is completed, an execution image of the virtual world may be represented as byte code. The byte code representing the virtual world may be executed by a runtime control to provide desired information to users.

Abstract: User Defined Routines (UDRs), capable of being expressed in one or more languages, are handled by determining a language native to the UDR, for example, by looking up a system catalog. If a language manager associated with the native language has not been loaded already, the language manager is loaded into a server memory. If the UDR has not already been instantiated, the UDR is instantiated and initialized. Then an execution context for the UDR is created and the UDR is executed. Loading of the language manager is handled by a general language interface capable of initializing the language manager, loading the language manager, creating a language manager context, and executing the language manager.

Abstract: Method and apparatus for optimizing the processing of join queries based on join cardinality. Embodiments implement the methods in query optimizers in relational database management systems. A good join order for a multiple join query is found with a metric that compares the relative merits of candidate join orders as a whole. Embodiments estimate the join selectivity of foreign key--foreign key joins, where both participating tables are foreign keys with respect to a primary or unique key of one primary table. A graph representation of a query is processed to estimate the join cardinality of an arbitrarily large number of filters and joins, including any combination of primary key--foreign key joins and foreign key--foreign key joins.

Abstract: A computer operated apparatus estimates values needed by an optimizer in a database management system (DBMS). The DBMS has one or more tables for storing data, each table having zero or more columns of user-definable data types and zero or more associated user-defined routines (UDRs). The apparatus has a feature vector extractor connected to the database tables for converting the UDR inputs into a base type representation. A neural network receives the feature vector and generates estimated values which are provided to the optimizer of the DBMS. The neural network can be trained periodically using randomly generated queries, or it can be trained dynamically by capturing data generated during a query. During operation, the optimizer dynamically invokes the neural network to generate estimates such as selectivity and cost per call for determining optimum query search sequence.

Abstract: A method and apparatus for estimating the number of distinct values in a database table selects one of several numerical methods to solve a model of the number of distinct values in a database table based on a table size, a sample size, and the number of distinct values in the sample. The sample may be a sample previously collected for purposes other than to estimate the number of distinct values, and the number of records in the sample and the number of distinct values in the sample may be determined at the time the sample was collected.

Abstract: User Defined Routines (UDRs), capable of being expressed in one or more languages, are handled by determining a language native to the UDR, for example, by looking up a system catalog. If a language manager associated with the native language has not been loaded already, the language manager is loaded into a server memory. If the UDR has not already been instantiated, the UDR is instantiated and initialized. Then an execution context for the UDR is created and the UDR is executed. Loading of the language manager is handled by a general language interface capable of initializing the language manager, loading the language manager, creating a language manager context, and executing the language manager.

Abstract: A computer implemented method for generating a response to a join. Two tables are divided into fragments. A join fragment map identifies, for each respective fragment from the first table, the fragments from the second table which contain a record that satisfies the join predicate with a record from the respective fragment from the first table. The map is used to eliminate fragments which cannot satisfy the join predicate.

Abstract: A computer-based method and system of processing records read respectively from first and second related tables of a database includes determining whether data contained in the records read from the first or second tables is unevenly distributed. Records corresponding to evenly distributed data are searched for matches using a first set of instances of the hash-join operator, and records corresponding to data which is unevenly distributed are searched for matches using a second set of instances of a hash-join operator.

Abstract: Information in a database is accessed with a computer system by transforming a file system request from an application into a database query and retrieving information corresponding to the database query from the database. The retrieved information is made available to the application as a file system object, for example, as a directory, a file, a link or a collection thereof.

Abstract: Routines for manipulating data in a database are performed by a database server by determining whether or not a database manipulation routine is to be isolated from a specified range of memory, and, in response, selectively switching memory access permission to the specified range of memory (e.g., from read/write to read-only) for the database manipulation routine under consideration. The database manipulation routine is then executed using the selectively switched memory access permission. A database manipulation routine that is to be isolated can be executed with read-only memory access thereby protecting the specified range of memory (e.g., corresponding to core data structures) from improper modification.

Abstract: In a computer-implemented method for generating a response to an aggregate query request on a source database, a random sample of records is extracted from the source database to generate a sample database. Then an aggregate query is performed on the sample database to generate a sample result, and an estimate of the aggregate query request on the source database is derived from the sample result.

Abstract: A computer-based method and system of processing records read respectively from first and second related tables of a database includes determining whether data contained in the records read from the first or second tables is unevenly distributed. Records corresponding to evenly distributed data are searched for matches using a first set of instances of the hash-join operator, and records corresponding to data which is unevenly distributed are searched for matches using a second set of instances of a hash-join operator.

Abstract: Generalized key indexes enable a first table of a relational database to be indexed using an index key and index conditions, wherein either or both of the index key and index conditions may reference multiple tables of the database or may be an expression using fields of one or more tables other than the first table. The generalized key indexes directly associate generalized index key values with record identifiers of records of the first table, thus enabling efficient storage and data retrieval.

Abstract: A computer writes information to a series of blocks that form a stripe on a disk array. The computer first checks if the stripe contains data written in a current pass. If the stripe has stored data, but is not completely full, the computer calculates a parity for the stripe (assuming the information is appended to the data). If appending the information to the data would not complete the stripe, the computer caches the parity outside the disk array. The computer appends the information to the data and updates a parity block on the stripe each time information is appended to the stripe. In various implementations, the computer places a stamp on the pages forming the blocks in the stripe to which information is written. Alternatively, the computer zeroes the bits in the remaining blocks to form a complete stripe.

Abstract: A computer-implemented method for making a data item available to a user program and a source program.

Abstract: A method and apparatus for providing per-instance data memory in a dynamic link library (DLL) linked to multiple tasks or to multiple instances of one task in the context of an operating system that does not provide such a feature. The invention includes providing a task switch enhancer DLL that allocates and manages a memory segment to store per-instance data of a DLL for a task as a substitute for an original per-instance data segment for that task.