International Business Machines Corporation

David L Levish

Stephanie H Lee

Christina Marie Lee

Michael R Shadduck

Ronald Elliott Parrish

W James Kincaid

Robert J Hollines, Iii

Stephanie Hui-An Lee

Jeffrey Allen Berger

Cecilia Carranza Lewis

San Yu Phoenix

Thomas Alexander Mahon

Yufen Wang

David W Slinger

Log records are generated by a computer system and stored in a computer readable storage area for recording changes to database objects responsive to receiving transaction requests, wherein the locators for the log records are of a predetermined maximum size that provides a maximum trackable size limit of the storage area. A computer system estimates an amount of time for the storing of the generated log records to fill the storage area up to the maximum trackable size limit. A computer system sets a soft limit to the storage area that is smaller than the maximum trackable size limit, including setting the soft limit based on the estimated amount of time.

Maintaining database consistency when nearing the end of a database recovery log

Provided are techniques for using a series of compare and swap operations to assign a Relative Byte Address (RBA) and a Logical Record Sequence Number (LRSN) to a log record. One or more log buffers for storing the log record are identified, and the log record is moved to the one or more log buffers.

Creating and maintaining order of a log stream

Embodiments of the invention comprise a method for dynamically switching a serialization method of a data structure. If use of the serialization mechanism is desired, an instruction to obtain the serialization mechanism is received. If use of the serialization mechanism is not desired and if the serialization mechanism is in use, an instruction to obtain the serialization mechanism is received. If use of the serialization mechanism is not desired and if the serialization mechanism is not in use, an instruction to access the data structure without obtaining the serialization mechanism is received.

Dynamically switching the serialization method of a data structure

The present invention provides a system and method for dynamic management of buffers by the buffer manager for a hierarchical file system. A system according to the present invention includes a buffer pool which is comprised of both fixed and virtual storage. The buffer pool is logically partitioned into three states, fixed, pageable and released, wherein each of the plurality of buffers in the buffer pool exists in one of these states. The system further comprises the buffer manager, which dynamically manages the buffers in the system by moving them between these three states in a non-disruptive manner based on a comparison between the actual usage of fixed and virtual storage compared with target usage values for the fixed and virtual storage.

System and method for performing dynamic buffer management in an HFS (hierarchical file system DF/SMS)

A method, system and program for accessing records in a database object, such as an index or table. A cursor is maintained that points to a start entry in the database object. The database object has rows and columns of values. A runtime component receives a command to return a kth entry in the object having column values that satisfy predicates in a query statement. The runtime component calls a manager component that accesses entries in the database object to return the kth entry in the object having column values that satisfy the predicates. The manager component moves the cursor through sequential entries in the database object, in either the forward or reverse direction, until a kth entry that satisfies the predicates from the start entry is reached. The manager component positions the cursor to an entry that is the kth entry that satisfies the predicates from the start entry and returns data from the kth entry pointed to by the cursor to the runtime component.

Method, system and program for scrolling index scans

Provided is a system, method, and program for performing a reverse scan of an index implemented as a tree of pages. Each leaf page includes one or more ordered index keys and previous and next pointers to the previous and next pages, respectively. The scan is searching for keys in the leaf pages that satisfy the search criteria. If a current index key is a first key on a current page, then a request is made for a conditional shared latch on a previous page prior to the current page. If the requested conditional shared latch is not granted, then the latch on the current page is released and a request is made for unconditional latches on the previous page and the current page. After receiving the latches on the previous and current pages, a determination is made of whether the current index key is on the current page if the current page was modified since the unconditional latch was requested. The current index key is located on the current page if the current index key is on the current page. A determination is then made of whether the located current index key satisfies the search criteria.

Method, system, and program for reverse index scanning

The method of the invention updates a tree arranged index for an hierarchical file system (HFS). The index includes at least one index value, i.e., an index page, that is logically positioned between an index anchor value and a sequence page. The sequence page includes actual index data. Logical positioning of the index page is determined by one or more pointers. The method initially responds to an index update request by making an update &ldquo;shadow&rdquo; copy of (i) a sequence page and (ii) any other index pages, up to and including a root page, that are to be updated in accord with the update request. Thereafter, an index manager updates the shadow copy of the sequence page in accord with the update request. The index manager further updates the root page and each shadowed index page that is present in a path to the updated sequence page to indicate that the path has been updated and includes the most current data. If a cancel or error indication occurs prior to updating of the index path, the shadowed copies of the sequence page, index page and root page are released, enabling the system to return to the non-updated pages. Otherwise, the index manager updates the index anchor value to point to the updated root page, thereby indicating that the index update has been successfully accomplished. Periodically, the new index anchor value, sequence page, index and root pages are written to disk.

Growing community of inventors

San Jose, CA, United States of America

David L Levish