摘要:
A computer-implemented method for minimizing the amount of time to access current data in a database which may be stored wholly in a DASD-oriented external storage subsystem or partly in DASD and partly in a stored high-speed electronic store while maintaining coherency of the data with respect to multiple user systems.
摘要:
A computer-implemented method for minimizing the amount of time to access current data in a database which may be stored wholly in a DASD-oriented external storage subsystem or partly in DASD and partly in a stored high-speed electronic store while maintaining coherency of the data with respect to multiple user systems.
摘要:
A method for ensuring data coherence while detecting whether the locally cached copy of a data page is invalid and responsively refreshing the locally cached page from a Shared Electronic Store (SES) in a multisystem shared disk environment. Locally cached data pages may become invalid in a multisystem shared disk environment because of transactions executed by other systems for the common database. Thus, whenever a transaction in a database management system (DBMS) instance desires to read or update a record in a locally cached data page, the DBMS must first verify validity for the locally cached copy and, for stale or invalid copies, must re-read and re-register the latest version in the SES. This invention provides a procedure for the necessary verification and refreshing steps that relies on page latching to serialize the combination of local multiuser and global multisystem activities. The procedure of this invention supports both record and page locking granularities.
摘要:
An efficient procedure for determining the set of buffer pool database pages that must be externalized to stable storage and for scheduling their write I/O's before release of a committing transaction's locks. In a multisystem database management system (DBMS) with high-speed shared external storage (SES) environment, a DBMS instance may follow "force-at-commit" protocol for a database that has intersystem read/write interest or may alternatively follow a "no-force-at-commit" policy when operating with a database for which only one system has interest. By introducing the concept of a series of unique ordinal numbers (ORD#) for each database assigned to buffer pool data pages whenever the page state changes from clean to dirty within a single DBMS instance, this procedure maintains a transaction page list (TPL) for each transaction in ORD# order. Because the dirty page list (DPL) pages are queued in ORD# sequence and each TPL includes all modified pages associated with the transaction, a merge scan with DPL entries can be performed for each TPL at the commit time of the corresponding transaction. This searches the DPL in a single pass with minimal DPL latch contention. The TPL pages requiring destaging are distributed to two processing lists; a first list for immediate force to SES and a second list of "in-use" pages that are forced to SES if necessary upon release of existing process latches.
摘要:
In a multi-system data sharing complex, a database system can write updated pages to a shared electronic store for a fast write. Other database systems can obtain pages written to the shared store for further modification without the pages first being written to stable storage. However, pages are eventually written to the stable storage in a castout process. Recovery of a database from failure of the shared store is bounded by determination of a recovery boundary which, when applied to the union of database system transaction logs, establishes a point in front of which are found log records of modifications to pages which were in the shared store when it failed. These log records are applied to page versions obtained from stable storage to recover from failure of the shared store.
摘要:
A computer system and a relational database management system (RDMS) computer program product are described for interfacing a number of concurrently running database sessions with a large database in which a plurality of local temporary object database descriptors are used in lieu of a single global temporary object database descriptor where the global and each of the local temporary object database descriptors include identifiers for temporary objects, each of the local temporary object database descriptors is associated with a single database session and each of the global database descriptors is shared among multiple database sessions.
摘要:
Prefetching data using a piecemeal list prefetching method. This is achieved by various means, including building a plurality of data pages, sorting the plurality of data pages into sequential data pages and a list of non-sequential pages, prefetching the sequential data pages using a first prefetching technique, and prefetching the non-sequential list of data pages using a second prefetching technique.
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:
A method and system to manage temporary storage space of a relational database management system by receiving a data segment allocation request from an application, allocating a data segment to the application, marking the data segment as allocated in a space map page, and indicating a usage status of the data segment. The method also includes receiving a mass delete command from the application, indicating reception of the mass delete command in a table-related control block, and marking the data segment as unavailable in the space map page. The method may also include indicating reception of the mass delete commit command in the table-related control block to enable the application to reuse the data segment.
摘要:
When recovering data in a database, initially, a database log, which records activities related to a database, is scanned. One or more individual objects that are to be recovered to a target time with reference to a backup time are identified.