Abstract: The present invention discloses a technique for restoring a database in a computer. In accordance with the present invention, the database contains objects and is stored on a data storage device connected to the computer. After a system failure, a log file is read. The log file contains one or more modifications to the database objects. Each modification has an associated data page and time stamp or sequence number. The modifications are sorted by at least one predefined sorting key value. The sorted modifications are then grouped by database object. The sorted modifications are applied to each database object in parallel.

Abstract: A method, apparatus, and article of manufacture for a computer implemented query compiler. A query is executed in a computer, the query being performed by the computer to retrieve data from a database stored on a data storage device connected to the computer. The query is received containing at least one variable. At execution time, a value is determined for each variable in the query. A filter factor is estimated using the determined value for each variable. Then, an access path is determined for the query using the estimated filter factor.

Abstract: A method and apparatus to provide DBMS restart recovery that allows transactions to access data that does not have restart recovery work pending. Access to data requiring restart recovery work to be performed is restricted, and a transaction is denied access to this data. In another embodiment, the invention allows certain transactions to access restricted data that has restart recovery work pending. These transactions are transactions that do not require data consistency. Regardless of the embodiment, the invention allows full recovery to be completed concurrent to the processing of new transactions requiring access to the database. An amount of restart recovery processing may be postponed until after the DBMS has begun accepting new work requests.

Abstract: A method, apparatus, and article of manufacture for a computer implemented query compiler. A query is executed in a computer, the query being performed by the computer to retrieve data from a database stored on a data storage device connected to the computer. The query is received containing at least one variable. At execution time, a value is determined for each variable in the query. A filter factor is estimated using the determined value for each variable. Then, an access path is determined for the query using the estimated filter factor.

Abstract: A method, apparatus, and article of manufacture for an encoder for encoding members in a concept hierarchy. A query is executed in a computer. The query is performed by the computer to retrieve data from a database stored on a data storage device connected to the computer. Members in the database that are related by one or more concept hierarchies are encoded. Then, members in one of the concept hierarchies that are descendants of one or more selected concepts are identified based on the encoding.

Abstract: A relational data base management system includes a query processor that permits consideration of alternative user-specified access paths, which the processor will validate and incorporate into the query execution plan it otherwise generates. The optimizer of the query processor retrieves user-specified data that defines an access path for a query during bind operations. Any access path parameters that are invalid are replaced with processor-selected values.

Abstract: The present invention provides a system and method for utilizing the proximity of keys in sequential or near sequential index probes to avoid complete index tree traversal. Page information from three pages (LAST, PARENT and NEXT) are stored in separate information fields within an Index Lookaside Buffer. The LAST information field contains information on the most recent leaf page accessed during an index probe in a read key or an insert key operation, the PARENT information field contains information on the parent page of the most recently accessed leaf page described in the LAST information field, and the NEXT information field contains information on the most recent leaf page accessed during a fetch-next key or delete key operation.

Abstract: The present invention provides a system and method for utilizing the proximity of keys in sequential or near sequential index probes to avoid complete index tree traversal. Page information from three pages (LAST, PARENT and NEXT) are stored in separate information fields within an Index Lookaside Buffer. The LAST information field contains information on the most recent leaf page accessed during an index probe in a read key or an insert key operation, the PARENT information field contains information on the parent page of the most recently accessed leaf page described in the LAST information field, and the NEXT information field contains information on the most recent leaf page accessed during a fetch-next key or delete key operation.

Abstract: A system and method are provided for recovering a computerized database. During transaction processing of a database, at least one log range is tracked comprising update transaction log records. The log range has a begin value and an end value. The begin value corresponds to a log RBA location of a first access of the database by an updating transaction. The end value corresponds to a log RBA location at a commit or abort of a last updating transaction. The end value corresponds to a point when the database is designated as being read only (called pseudo open). During recovery of the database, the update transaction log records in the log range are applied to a version of the database stored in non-volatile memory.

Abstract: A system and method are provided for recovering a computerized database. During transaction processing of a database, at least one log range is tracked comprising update transaction log records. The log range has a begin value and an end value. The begin value corresponds to a log RBA location of a first access of the database by an updating transaction. The end value corresponds to a log RBA location at a commit or abort of a last updating transaction. The end value corresponds to a point when the database is designated as being read only (called pseudo open). During recovery of the database, the update transaction log records in the log range are applied to a version of the database stored in non-volatile memory.

Abstract: Database files containing records include pages called free space inventory pages (FSIPs) describing field space information relating to data pages. In a transaction processing system, the invention provides correct sequences for logging of updates to FSIPs when the updates are required by updates or UNDOs to data records. If, during operation to insert a data record to a data page, the FSIP containing free space information for the page indicates that the page is empty and there are no uncommitted deletes to the page, page I/O is avoided by formatting the page directly in a data buffer pool without reading the page from disk. During a cursor stability-level table scan with data record-level locking, excessive I/O and some record locking are avoided by using space reservation fields on an FSIP to ensure that there is no space reserved on the data page for a later undo of uncommitted data records deletes from the page.

Abstract: An automated method is provided for optimizing a query for retrieving data from a relational database in a computer relational database management system. The method uses rules for identifying queries for which duplicates are a factor in retrieving the proper data from the database. The method then merges queries with table expressions, views and subqueries when possible without causing duplicates to be unnecessarily retrieved.

Abstract: Results of a relational data base management system are joined in a process requiring, first, existence of an index on the join columns of the inner table, and, second, ordering on the join column of the first table. First, the index on the inner table's join column is scanned for rows of the inner table having join column values matching such values of rows in the outer table. This is done in a single pass through the outer table. Next, a temporary work table containing the identifiers of inner table rows having join column values matching those of the outer table is produced by concatenating the row identifiers to their matching outer table rows. Following this, the temporary work table is ordered by the identifiers. Last, the identifier list of inner table rows is used to retrieve the corresponding rows of the inner table.

Abstract: An implementation of referential integrity in which descriptions of referential constraints are compiled into meta-data descriptions of the constraint rules and specifications. The meta-data descriptions of the constraints are stored in the form of objects called relationship descriptors. Each relationship descriptor contains a complete description of a referential constraint, either directly or by means of pointers to other objects such as record and index descriptors which contain information comprised in the constraint's specification. The relationship descriptors are linked into two types of chains by symbolic pointers. One type of relationship descriptor chain connects all relationship descriptors which have a common parent table. The other type of relationship descriptor chain connects relationship descriptors with common dependent tables. Both types of chains are anchored in respective fields in the tables' record descriptors.

Abstract: A page-accessing method in a segmented tablespace 10 which eliminates unnecessary reading and locking. The tablespace comprises data pages 18 grouped into identically-sized segments 16, each segment storing data for a single table. A status indicator 26 for each data page of a segment is kept in a separate segment control block 20 stored on a space map page 14. Five data page status indicator values are maintained:(1) FULL (26a) - entirely full of current data;(2) PARTIALLY FULL (26b) - partially full of current data;(3) UNFORMATTED (26c) - empty; contains no data;(4) MASS DELETE (26d) - contains only obsolete data because of an unqualified deletion (mass delete) of data; or(5) QUALIFIED DELETE (26e) - contains only obsolete data because of a qualified deletion of data.When scanning over the data in a segment, UNFORMATTED and MASS DELETE pages are skipped. QUALIFIED DELETE pages are locked, and then skipped if they still contain only obsolete data when the lock is obtained.

Abstract: A method is disclosed for enforcing referential constraints on a record-by-record basis, immediately before or after each record is manipulated and while the record is still accessed, significantly improving the system's performance. Each record is visited only once to do both the constraint checking and the manipulation (insert/update/delete). If the constraint checking fails, then the entire relational operation of which the record manipulation is a part is backed out. For insertions, each record is first inserted 30, and then constraints respective the record are enforced 32,34. For updates, the record is updated 48 after constraints respecting the record's primary key are enforced 40,42, and before constraints respecting its foreign key(s) are enforced 44,46. Deletions are performed 52 before the constraints on the deleted record are enforced 54,56. Cascade deleted 58 are handled recursively 60. The method correctly processes cyclic constraints and self-referencing rows without special handling.

Abstract: A method is disclosed for deferring enforcement of referential constraints in large-scale data base operations such as the population or loading of relational tables 10, 12. First, the new rows are loaded in a Data Load phase 24 into the tables, and information on the new rows and their constraints is extracted and sorted 26 into a sorted key data set 68. Any primary indexes 22 required for constraint checking are then updated 28 using the sorted key data set. The new rows are then checked 70 for constraint violations, such violations are rectified 72 to restore the table's referential integrity, and a deletion data set 80 is produced. The deletion data set is merged and sorted 86 with row information 53 stored during loading, and the result is used to copy 88 the new rows violating referential constraints to a separate discard data set 48 where they can be corrected and reloaded into the table. Finally, an error summary report 36 is generated for use in correcting the discard data set.

Abstract: A method utilizing a coordinated pair of locking limits for managing concurrency and lock granularity tradeoff relations. A first limit is placed on the number of small granularity locks per resource. A second limit is placed on the number of locks assignable to each process. When the first number of small locks is reached, the method withdraws the small locks and grants only one lock to the entire resource (lock escalation). When a process requests an additional lock over the second limit, the lock is refused.

Abstract: A method for assuring atomicity of user requested multi-row update operations to tables such as in a relational database, guarantees that for any update operation that succeeds all stated effects will have occurred and that for any update operation that fails the system state as perceived by the user remains unchanged. This is accomplished by establishing, in response to a multi-row update operation request, an execution module of a program containing sets of machine language code instructions implementing the update operation request with a savepoint request at the beginning of the execution module of the program. For each set of machine language code instructions in or called by the execution module which modified the user perceived system state, information is logged to a soft log.