Abstract: A method and apparatus for selective removal of user data changes is provided. In one embodiment, the methodology is implemented by a logical recovery mechanism. Upon receiving a request to selectively remove user data changes stored in a database, the mechanism creates an empty set of transactions to be populated for selective removal, identifies one or more transactions and places them in the causal set of transactions. The mechanism further identifies other transactions that logically depend on any transaction already in the causal set of transactions. Finally, either while the causal set of transactions is being identified or after the identification is completed, the mechanism also determines user data changes made by the causal set of transactions. One or more compensating transactions may be generated by the logical recovery mechanism, which when executed, reverse the user data changes.

Abstract: A method, system, and computer program product is disclosed for caching results in a client-side cache. Embodiments of a method, a system, and a computer program product are disclosed that associate a first snapshot of a database with a client that indicates a state of the database after a last database request by the client, and indicate any number of invalid cached results in the client cache for the client based upon the first snapshot. In some embodiments, the method further includes receiving a second snapshot that indicates a state of the database upon receipt of a database server request by the client, and updating the first snapshot with the second snapshot.

Abstract: A technique for making versions of rows of a user-defined tracked table temporally queryable. The database management system of the technique permits temporal queries of user-defined tables. The queries return versions of rows in the user-defined table that are currently in an undo log maintained by the database system. Associated with the tracked table are a system history table which contains versions of the rows and temporal metadata indicating when the versions were in the tracked table and a system form history table which contains versions of the form of the tracked table and metadata indicating when the tracked table had the form. These tables are created and maintained by a background process in the database management system. A temporal query on the tracked table combines results from the undo log with results from the system history table. The results' form is determined by the system form history table.

Abstract: A method, system, and computer program product is disclosed for interacting with a client supported by a client-side cache. Embodiments of a method, a system, and a computer program product are disclosed that retrieve a first snapshot, indicating a state of the database after a last database request by the client, associated with the client, determine any number of invalid cached results for the client based on the first snapshot, and transmit the any number of invalid cached results and a second snapshot, an update for the first snapshot.

Abstract: Described herein is a session reuse approach that reuses a session already established on a remote participant to execute an autonomous transaction. The session is reused in a manner that avoids affecting the session's state in a way adverse to later execution of the containing transaction. When beginning an autonomous transaction on a local participant, a request to initiate an autonomous transaction on a remote participant is deferred or not sent at all if the containing transaction is a distributed one. The request may be sent later, piggybacked to another message, using a piggyback messaging system. When ending an autonomous transaction, a local participant may send a request to a remote participant requesting that the remote participant commence execution of its respective containing transaction. The request is piggybacked to another message, and is thus transmitted without having to transmit the request separately, avoiding a message round trip.

Abstract: A bulk delete operation with reduced latency allows for retrieval of deleted data. Each database object holds a Delete SCN indicating when a bulk delete operation was last performed. Each row in the database object holds a Row Modification SCN indicating when the row was last updated. A bulk delete is performed by writing the old value of the Delete SCN to the undo tablespace and updating the Delete SCN. No undo information is stored for the rows. A write is performed by finding a deleted row, storing undo information for the deleted row and writing over the deleted row. To read from the database object, a rollback operation is performed, if necessary. Those rows are then retrieved for which the Row Modification SCN is higher than the Delete SCN and is less than or equal to the timestamp for the requested data.

Abstract: Techniques are provided for providing mappings between logical time values and real time values in a multinode system. The techniques include, if a first event occurs, then writing an LTV-to-RTV mapping to a non-volatile data structure. If a second event occurs, then a check is performed to determine whether any of the LTV-to-RTV mappings in the volatile data structure are newer than all LTV-to-RTV mappings in a non-volatile data structure. If there are newer mappings in the volatile data structure, then one or more LTV-to-RTV mappings are written to the non-volatile data structure based on those “newer” LTV-to-RTV mappings from the volatile data structure. Upon receiving a request an LTV for a particular RTV, the particular RTV is determined based on information from an information source such as the volatile data structure or the non-volatile data structure.

Abstract: Techniques are provided for providing mappings between logical time values and real time values for a database. The techniques include, if a first event related to the database occurs, writing an entry of a current LTV and a current RTV in a volatile mapping data structure; and if a second event related to the database occurs, writing one or more entries in a non-volatile mapping data structure based on information contained in the volatile mapping data structure. The techniques also include, once a request from a requestor to provide an LTV that is mapped to a particular RTV is received, determining the LTV for the particular RTV, based on the particular RTV and information from a source, wherein the source is one of the volatile mapping data structure and the non-volatile mapping data structure; and providing the LTV that is mapped to the particular RTV to the requestor.

Abstract: Methods and systems for automatically tuning the undo retention in a database system are disclosed. The undo tablespace may be of a fixed size or may be set to be automatically extended. If the undo tablespace is of a fixed size, the undo retention is set to a “best possible” undo retention, which is calculated based on statistical data stored in the system. If the undo tablespace is configured to be automatically extensible, the undo retention may be set to be greater than the duration of the longest running query in the system. The duration of the longest running query may be calculated in a distributed process in which each instance broadcasts information about its longest running query to other instances.

Abstract: A method and apparatus for database-driven distributed recovery is provided. According to one aspect, a transaction manager sends, to one or more resource managers, connection information that indicates how to connect to the transaction manager. A resource manager that receives the connection information establishes an association between the connection information and the distributed transaction for which the connection information was received. If the resource manager is ever in doubt concerning whether the distributed transaction should be committed, the resource manager uses the associated connection information to send, to the transaction manager, a request for status information. The transaction manager maintains status information that indicates whether the distributed transaction should be committed, and responds to the resource manager with the requested status information.

Abstract: A method and apparatus for performing file system operation locks at a database server is provided. A request to perform a file operation on a portion of a file managed by a database server is received at the database server. In response to receiving the request, the database server grants a lock that covers only a portion of the file that is involved in the file operation. For example, the database server may grant a lock that covers a range of bytes on the file, where the range of bytes is less than the entire file. Thereafter, the database server performs the file operation on the file. The file operation may be a NFS operation.

Abstract: Embodiments of the present invention include a method for performing a data uniqueness check in a set of sorted data blocks comprising accessing data to be inserted into the set of sorted data blocks, determining a proper data block of the set of sorted data blocks for which the data to be inserted should be located based on a first sorting key. The method further includes examining the proper data block for an existing data entry and if the proper data block is empty, the method further includes inserting the data to be inserted into the proper data block of the set of sorted data blocks.

Abstract: Techniques are provided for sharing of flashback cursors by adding a time domain property to flashback cursors. This time domain property defines the range of flashback times for which the flashback cursor is valid. According to one embodiment, this “validity range” is closed at the lower bound and open at the upper bound. A subsequent flashback query can share an existing flashback cursor if the flashback time of the subsequent flashback query falls within the validity range of the existing flashback cursor. In one embodiment, the validity range of a flashback cursor is established based on times associated with indexes used to process the flashback query for which the flashback cursor was made. Consequently, an existing flashback cursor is less likely to be used by a subsequent flashback query when it would be inefficient to do so.

Abstract: A method and mechanism is disclosed for implementing transaction logging in a database system. In-memory undo records are maintained to log undo information for the database system. Redo records are batch processed, with multiple redo records for a transaction stored on disk at commit time.

Abstract: Techniques are provided for providing mappings between logical time values and real time values in a multinode system. The techniques include, if a first event occurs, then writing an LTV-to-RTV mapping to a non-volatile data structure. If a second event occurs, then a check is performed to determine whether any of the LTV-to-RTV mappings in the volatile data structure are newer than all LTV-to-RTV mappings in a non-volatile data structure. If there are newer mappings in the volatile data structure, then one or more LTV-to-RTV mappings are written to the non-volatile data structure based on those “newer” LTV-to-RTV mappings from the volatile data structure. Upon receiving a request an LTV for a particular RTV, the particular RTV is determined based on information from an information source such as the volatile data structure or the non-volatile data structure.

Abstract: Techniques are provided for providing mappings between logical time values and real time values for a database. The techniques include, if a first event related to the database occurs, writing an entry of a current LTV and a current RTV in a volatile mapping data structure; and if a second event related to the database occurs, writing one or more entries in a non-volatile mapping data structure based on information contained in the volatile mapping data structure. The techniques also include, once a request from a requestor to provide an LTV that is mapped to a particular RTV is received, determining the LTV for the particular RTV, based on the particular RTV and information from a source, wherein the source is one of the volatile mapping data structure and the non-volatile mapping data structure; and providing the LTV that is mapped to the particular RTV to the requestor.

Abstract: Processing a database statement received from a database client, which requests information about the timing of a last committed change to a row in a database table, comprises retrieving a logical time value that represents a commit time associated with the row and providing the value to the client. In one embodiment, providing the logical time value to the client is enabled by database infrastructure in which a virtual column is used to temporarily “store” the value so that the value can be manipulated as if the value resided in a column of the row. Performance of operations that involve the virtual column are processed similar to real columns, as if data was actually logically stored in the virtual column. For example, the virtual column can be used in condition clauses of database statements for concurrency control.

Abstract: A method and apparatus for data recovery are disclosed. Undo tablespace size is calculated for user-specified undo retention time based on system statistics collected over a period of time specified by a history time parameter.

Abstract: A method and apparatus for creating a target database on a target platform based on a source database on a source platform is provided. A target database is created on the target platform based on data, metadata (for example, data in the system tablespace and control files), and external objects (for example, password files and externally stored tables) comprised within the source database. Data and metadata stored in tablespaces of the source database are converted to the format compatible with the target platform. The conversion of the data and metadata may be performed either at the source database or the target database. Redo information and undo information of the source database are not copied from the source database to the target database. Certain files, e.g., a password files, and links to externally stored object may be recreated on the target database.

Abstract: Disclosed is a mechanism, system, interface, and process to determine one or more life intervals when given a key. In addition, one or more life intervals can be determined for each key within a range of keys. This type of process is useful for auditing purposes or if it is desire to implement a query upon past versions when a key or range of keys is specified for the data for which past versions is desired. This facilitates algorithms, mechanisms, and processes to implement a query upon past versions.