摘要:
According to the present invention a method and a system is provided for establishing synchronized recovery log points in an infrastructure comprising at least a first database management system and a second database management system, whereby recovery log files are written and log points are created separately for each database management system. Firstly, log write of all database management systems is suspended, then, the corresponding log points for each database management system forming a global log point are recorded and log writes of all database management systems are resumed.
摘要:
According to the method, apparatus, and computer readable medium of the present invention, a scan is performed by accessing all rows of a table in a relational database, evaluating each row to determine whether the row satifies a set of predicates of a database query, and returning the row if it satisfies the set of predicates. According to a preferred embodiment of the present invention, all rows are evaluated regardless of current locks. The scan is continued if the row does not satisfy the set of predicates regardless of the current lock on the row. The locking semantics according to the invention has the advantage that it completely removes the problem of lock contention on the rows that do not satisfy statement predicates.
摘要:
In a database management system (DBMS) (60) for a database application (10) including a database (12) having a table (14, 16, 18) and a unique key index (42) having indexes (44, 46, 48) therefor, the DBMS (60) includes a data manager (64), an index manager (66), a transaction manager (62), and a lock manager (68) which restricts access to the table by assigning locks to elements thereof. In order to avoid deadlock in the database application due to the pseudo-deleted entries, the lock categories include an X-lock and a Conditional S-lock, and have lock attributes including at least a Delete attribute for the X-lock. The Conditional S-lock is compatible (granted by the lock manager) with an X-lock whose Delete attribute is NOT SET, but is not compatible (granted) with an X-lock whose Delete attribute is SET. Each index entry includes a pseudo-delete flag which is SET by the index manager (66) to indicate deletion of the indexed row. Conditional upon locating a pseudo-deleted index key corresponding to a key to be added to the index, the index manager (66) requests a Conditional S-lock on the row indexed thereby, whereby the index manager (66) verifies the Delete transaction which set the pseudo-delete flag has committed.
摘要:
A system for retrieving stored data and/or index pages, receives a database request, generates an access plan for the database request for accessing data rows and/or index entries according to the statement, and generates a priming process by selecting a set of data and/or index pages. The set of data and/or index pages is a super-set or the exact set of the pages needed for accessing data rows and/or index entries according to the access plan. The present system retrieves the data and/or index pages of the set before executing the database request according to the access plan. The execution of the database request is performed by scanning the index and data pages preloaded into the cache. This reduces the number of necessary disk reads during execution.
摘要:
The present system indexes a plurality of entries in a database that contains a database table having a base index. As a recent row is inserted in the database table, an index increment is generated based on the inserted row. Preferably, the index increment is smaller in size than the base index because it is recently generated. The smaller size of the index increment facilitates the management of the index increment. An index entry associated with the inserted row is added to the index increment, and the index increment is merged with the base index.
摘要:
A method for performing a system level backup of a log-ahead database management system (DBMS) without suspending updates by application programs is described. The DBMS comprises a database mainline system, a backup utility and a restore utility. The data and log records are stored on separate storage volumes. Log records are written to identify objects that require special handling during the point-in-time recovery. The database engine operates normally during a backup except for suspending actions that would alter the file system catalog or write updates across a storage volume boundary; and by freezing the REDO log point in its checkpoint information. The backup utility copies the data volumes first and optionally the log volumes second while updates are allowed. The resulting inconsistencies are resolved either during a DBMS restart or during a point-in-time (PIT) recovery performed by the restore utility.
摘要:
A computer program product provides accurate statistics in real time. These statistics can be used to determine if a table space in a database management system (DBMS) requires maintenance operations such as reorganization, back up, fresh access path statistics, and/or larger disk space allocation. The DBMS maintains a set of values that provide indications of whether that operation is due on that object. The indicators are asynchronously externalized in dedicated database tables either periodically in user-specified intervals or at some predefined event such as stopping the database. When the user performs a database administration operation, the database management system resets the associated indicators, and reports objects that are in an exception state based on the indicators collected as described for each operation.
摘要:
A method and system for formatting space in a storage device for a database system is disclosed. The method and system include formatting a first increment for creating a database file and asynchronously formatting at least one subsequent increment for the database file in a background of the database system.
摘要:
Methods and apparatus, including computer program products, for maintaining a set of indexes in a database management system (DBMS) having at least one table. A current, stale or deferred status is defined for at least a part of the indexes, resulting in at least a part of a set of current, stale, or deferred indexes in the DBMS. Current indexes are maintained by refreshing a current index synchronously with a table change relating to the current index. Stale indexes are maintained by refreshing a stale index continuously and asynchronously to table modifications of tables relating to the stale index based on log information relating to the modifications. Deferred indexes are maintained by building a deferred index in response to a query to a table relating to the deferred index, thereby bringing the deferred index in accordance with the current query time status to the table relating to the deferred index.
摘要:
According to an embodiment of the present invention, a method for reorganizing a source index tree of a database table resulting in a target index tree of the database table is provided. A backup process reads source leaf pages of the source index tree in a logical order defined by a sequence of index keys. The logical order can be different from a physical order of the source leaf pages defined by a sequence of respective storage locations. The backup process constructs target leaf pages of the target index tree maintaining the logical order. A restore process constructs target non-leaf pages of the target index tree based on the distribution of the index keys among the constructed leaf pages and stores the target leaf and non-leaf pages.