International Business Machines Corporation

Robert William Lyle

Ping Wang

Julie Ann Watts

Jay A Yothers

Christina Marie Lee

Balakrishna Raghavendra Iyer

James Z Teng

James Zu-Chia Teng

Charles Roy Bonner

Karelle Lynne Cornwell

Kevin Scott Beyer

Oliver Draese

Edison Lao Ting

Michael Jeffrey Winer

Donald J Haderle

Kalpana Shyam

Tanya Couch

Akira Shibamiya

Jerry Mukai

Steven John Watts

Atul Chadha

George F Wang

Hsiao Ming Wang

Shirley Zhou

Wei Ey Li

Curt Lee Cotner

Matthew Albert Huras

Ian R Finlay

Eugene Jon Shekita

Sridhar Rajagopalan

Mengchu Cai

Ruiping Li

Aamer U Sachedina

Yao Ching Stephen Chen

Michael R Shadduck

Robert J Hollines, Iii

Yufen Wang

Stephen J Schmandt

Krishnarao G Kulkarni

Barbara Tanya Couch

A database-management system evaluates a query that retrieves and transforms encoded symbols stored in a database. If the stored symbols assume a relatively small set of distinct values, the system initially performs the transformation on every value in the set. During execution of subsequent queries, rather than performing the transformation upon every stored symbol fetched from the database, the system merely returns the previously derived encoded transformation results that correspond to the decoded value of each fetched symbol. If the symbols stored in the database span a relatively large set of distinct values, the system does not initially perform the transformation upon every value in the set. Instead, the first time the system fetches a symbol that has a particular value, it saves that symbol's encoded transformation result and reuses that result the next time it fetches an encoded symbol with the same value.

Derived data dictionary for optimizing transformations of encoded data

Evaluating LIKE predicates against compressed data. An alphabet, a LIKE predicate, a compressed string, and a compression dictionary for the compressed string are received. Entries in the compression dictionary include a character string and an associated token. The LIKE predicate is converted to an equivalent pattern matching form, involving a search pattern of length m. For each character of the alphabet that appears in a string associated with a token, a mask of predetermined length k is created. For each entry in the compression dictionary a cumulative mask of length k is computed. A bit vector of length k is initialized, based on the search pattern. Successive tokens in the compressed string are processed using a logical shift of the bit vector and a bitwise operation of the bit vector with the cumulative mask associated with the token.

Fast evaluation of predicates against compressed data

In a high insert throughput in a database system, the database management system (DBMS) caches, asynchronously from data insert transactions, a list of storage location identifiers for storage locations in a table space in the database system. When a request for an insert transaction with an insert size is received from an agent, the DBMS identifies a storage location on the list with an amount of free space meeting the insert size. The DBMS removes the storage location identifier from the list and returns the storage location identifier for the storage location. When the insert transaction completes, the DBMS determines a remaining free space at the storage location and adds the storage location identifier to the list according to the remaining free space. By removing storage location identifiers from the list, another agent cannot receive the same storage location identifier for a parallel task. Thus, space contention is avoided.

Method and technique of achieving extraordinarily high insert throughput

Techniques for decompressing data in a database system. A query is received, which pertains to a subset of data within a compressed set of data. One or more decompression strategies are evaluated using a cost model. The cost model includes an estimated filter factor. A low cost decompression strategy is selected based on the results of the evaluation of the one or more decompression strategies. One or more bytes representing the requested subset of data are located within the compressed set of data. Only a portion of the compressed data that corresponds to the subset of data is decompressed, using the selected decompression strategy, while leaving the remaining set of data in a compressed state.

Data access performance using decompression maps

Methods and apparatus, including computer program products, implementing and using techniques for decompressing data in a database system. A query is received, which pertains to a subset of data within a compressed set of data. One or more decompression strategies are evaluated using a cost model. The cost model includes an estimated filter factor. A low cost decompression strategy is selected based on the results of the evaluation of the one or more decompression strategies. One or more bytes representing the requested subset of data are located within the compressed set of data. Only a portion of the compressed data that corresponds to the subset of data is decompressed, using the selected decompression strategy, while leaving the remaining set of data in a compressed state.

In a remote index operation, a first system in a datasharing group performs an operation on a data object in a database, determines a second system in the datasharing group has exclusive access to an index page to be updated according to the operation, and sends to the second system a remote request to change the index page according to the operation. In response, the second system changes the index page with an index entry referencing the data object and marks a key in the index entry as a provisional key. When a third system in the datasharing group reads the index entry, it determines that the key in the index entry is the provisional key. In response, the third system determines whether the data object exists in the database and a committed state of the transaction, and accordingly determines a current state of the data object.

Alleviation of index hot spots in data sharing environment with remote update and provisional keys

Embodiments provide an approach for efficiently searching a temporal key index as well as enforcing key distinctiveness property in a database table corresponding to the index. For example, by addressing time range overlaps in an index having a sort order for temporal keys, a database management tool of a server system may efficiently process client requests while enforcing a temporal key property in the database table.

Managing a temporal key property in a database management system

Optimizing approximation of an earliest access time of objects in a datasharing environment by maintaining a global exception list for objects that are accessed by transactions that commit infrequently. The global exception list comprises a time value representing the earliest access time for each object in the global exception list, and an object identifier. The system references the global exception list to determine whether the object is in the global exception list. If the object is in the global exception list, the system uses the time value associated with the object when determining the earliest access time for the object. If not, the system approximates the earliest access time of the object by selecting the most recent access time value from the global exception list, providing more recent access times for objects accessed by transactions that commit frequently.

Efficient tracking of an alternate global update and read values using exception lists

In searching an index in an original search structure for a database, portion(s) of the index are determined to be stored in a memory in an alternate search structure. The alternate search structure with the portion(s) is created and stored in the memory. The index is searched by traversing the alternate search structure and possibly the original search structure. The alternate search structure may be traversed more efficiently than the original search structure. The number of portions of the index may be a tunable parameter or selected based on a performance improvement/expense ratio. The search of the index begins at the alternate search structure. During traversal of the one or more portions in the alternate search structure, if traversal is to continue at a portion in the original search structure, the identity of this portion is obtained. The traversal is continued at this identified portion in the original search structure.

Index traversals utilizing alternate in-memory search structure and system memory costing

Index traversals utilizing alternative in-memory search structure and system memory costing

Alleviation of index hot spots in datasharing environment with remote update and provisional keys

A system and method for managing and storing logically grouped hierarchical data via physical block storage is provided. Logical groups of parsed XML node data forming node ID ranges are indexed by creating and inserting an index entry into a node ID range. Index entries indicate node ID range bounds for blocks in which nodes are stored. Consulting a node ID range index facilitates XML node traversal via logical links between nodes in different blocks. Additionally, physical links between nodes within a block allow for fast node traversal. Node update including insertion and deletion as well as document order based pre-fetch and XML document re-organization is also facilitated by this architecture.

Hierarchical storage architecture using node ID ranges

Growing community of inventors

Morgan Hill, CA, United States of America

Robert William Lyle