International Business Machines Corporation

Unisys Corporation

Lee-Chin Hsu Liu

Yun Yu Wang

Yao-Ching Stephen Chen

Guogen Zhang

You-Chin Gene Fuh

Fen-Ling Lin

James M Plasek

Ruiping Li

Hong Sang Tie

Li-Mey P Lee

Allan Benjamin Lebovitz

Irene Ching-Hua Liu

Binghua Zhen

Shyh-Yee Wang

Kenji Yoneda

Patrick Dooling Bossman

Ralph E Sipple

Jerry Mukai

Thomas P Cooper

Brian Thinh-Vinh Tran

Christina Marie Lee

Yumi Kimura Tsuji

Jennifer J Smith

John R Jordan

Ronald Q Smith

Desai Paramesh Sampatrai

Michael S Jende

Ding-Wei Chieh

James A Elke

Lester Y Lee

Michael J Bierma

James R Mettling

Hsuiying Yen Cheng

Rajeev P Tandon

Katsuhiko Kihara

Sheng-te Chiang

Lowell L Baerenwald

A system and computer readable medium for executing a query to access data stored in a database, wherein the query includes a plurality of DISTINCT keys, is disclosed. The system and computer readable medium includes a capture module for identifying each of the plurality of DISTINCT keys in the query and a sort module coupled to the capture module for determining if more than one sort is needed to execute the query, performing a first DISTINCT operation on a first DISTINCT key of the plurality of DISTINCT keys, storing data fetched from the first DISTINCT operation in a master workfile only if more than one sort process is needed to execute the query, and utilizing the master workfile to perform subsequent DISTINCT operations on the other of the plurality of DISTINCT keys.

System for executing a query having multiple distinct key columns

A method for executing a query to access data stored in a database, wherein the query includes a plurality of DISTINCT keys, is disclosed. The method of the present invention includes providing a first DISTINCT operation on a first DISTINCT key, storing data fetched from the first DISTINCT operation in a master workfile if more than one sort process is needed to execute the query, and utilizing the master workfile to perform subsequent DISTINCT operations for the other DISTINCT keys.

Method for executing a query having multiple distinct key columns

A method and system for efficiently identifying materialized view candidates for queries filters materialized views using certain criteria, using the materialized view signatures. This filtering rejects some of the unqualified materialized views prior to the performance of the query rewrite matching algorithm, resulting in a group of materialized view candidates. The query rewrite matching algorithm is then performed on the materialized view candidates. By first filtering the materialized views based on their signatures, the number of materialized views on which the query rewrite matching algorithm is performed is significantly reduced, improving performance.

Materialized view signature and efficient identification of materialized view candidates for queries

A method, apparatus, and article of manufacture for optimizing a query in a computer system, wherein the query is performed by the computer system to retrieve data from a database stored on the computer system. The optimization is performed by identifying a correlated subquery in the query that qualifies for transformation, transitively closing a local filtering predicate within the identified correlated subquery that is on a same column as a correlation join predicate within the identified correlated subquery, and pulling out or bubbling up the transitively closed predicate from the correlated subquery to a parent query block of the query.

Query transformation for queries involving correlated subqueries having correlation join predicates with local filtering predicates involving predicate transitive closure and predicate pull-out

A heuristic approach is used to order materialized view (MW) candidates in a list based on descending order of their reduction power. A query (e.g., SQL query) is then matched with the MVs in the list order, wherein searching is stopped when matching has been found. The query is matched with materialized views in the ordered list by identifying a materialized view candidate as follows: identifying an MV that is not locked by a REFRESH process; identifying a matching MV that does not require a regroup; identifying a matching MV that does not require a rejoin; identifying a matching MV that does not require a residual join; or identifying an MV with largest reduction power from the list of candidates.

Efficient heuristic approach in selection of materialized views when there are multiple matchings to an SQL query

Disclosed is a system, method, and program for searching a list of entries. Each entry is comprised of multiple fields. Each field in the entry may include one of a range of possible values for that field. If search criteria is provided for less than all of the fields, then multiple search keys will be generated such that each generated search key includes a search value for each field. The search value for each field for which there are no provided search criteria comprises one of the possible values for that field. A search operation is executed for each of the generated search keys to determine entries in the list that include field values that match the search values in the search key.

Method, system, and program for searching a list of entries when search criteria is provided for less than all of the fields in an entry

Disclosed is a system, method, and program for joining a multi-column table and at least two satellite tables. Each satellite table is comprised of multiple rows and at least one join column and each multi-column table is comprised of multiple rows and join columns corresponding to the join columns in the satellite tables. A query including predicates is received. A join predicate column comprises the satellite table and multi-column table join column to which at least one query predicate applies. A determination is then made as to whether there is at least one index on the multi-column table including at least one column for one join predicate column. One index is selected. The ordering of the join predicate columns in the selected index is used to determine the join order of the satellite tables and the multi-column table. The satellite tables and multi-column tables are then joined in the determined join order.

Method, system, and program for determining the join ordering of tables in a join query

A secondary index search in a relational database system compares primary key selection criteria against a primary key value stored in a secondary index record that satisfies secondary key selection criteria instead of searching a primary index for a primary key value that satisfies the primary key selection criteria.

Secondary index search

A hybrid hash join process joins data rows from two tables which have at least one common data column by partitioning the data rows based on the values in the common data column(s), creating data structures to decrease search time for matching rows, and recovering full data buffers using a unique buffer management methodology. A smaller one of the two tables is designated as an outer table and a larger one of the two tables as an inner table. The hybrid hash join process determines which rows in the inner and outer tables satisfy a selection criteria; the rows that satisfy the selection criteria are referred to as inner hit rows and outer hit rows. The hybrid hash join process assigns the inner and outer hit rows to corresponding inner and outer partitions, respectively. Buffer overflow in the outer partitions is handled by linking empty buffers to the outer partitions until all buffers are used. Buffer space is recovered by writing outer hit rows for a selected partition to mass storage. Buffer overflow in the inner partitions is handled by either writing the inner rows in the buffer to mass storage or by searching the corresponding outer partition for matches. The outer hit rows in the corresponding outer partition are allocated to entries in a data structure which is then probed to find matches for the inner hit rows in the buffer. Matching outer hit rows for any inner hit rows written to mass storage are found by repeatedly reading into memory inner and outer hit rows and probing a data structure created from the memory-resident outer hit rows in a partition. Optionally, binary trees are built from the entries in the data structure to speed up the probing procedure. The hybrid hash join process uses at least one hashing algorithm to assign hit rows to partitions, to allocate rows to entries in the data structures, and to probe the data structures for matches.

Hybrid hash join process

A merge join process combines rows from an inner and an outer table when the inner table is indexed on a data column that is common to both tables. The merge join process creates a set of rows from the outer table that satisfy a selection criteria and sorts the rows in the set on the common data column if necessary. The merge join process searches for a matching inner row for each outer row in sequence using the inner table indices until it finds a matching inner row on a data page. The data page is then repeatedly searched for matches on the successive outer rows in the set until the end of the data page is reached. The end of the data page is reached when the value of the common data column in an outer row is greater than a last key that represents the highest value of the common data column in a inner row stored on the data page. The merge join process also determines that a matching inner row does not exist in the inner table when the value of the common data column in an outer row is less than a next key that represents the lowest value of the common data column that has not yet been searched for a match. Thus, the next and last keys reduce the number of scans of the data page. All matching inner rows on the data page are marked as join rows; all outer rows with common data column values between the next and last keys but which did not have matching inner rows are discarded. When the end of the data page is reached, another data page is located and the search is repeated. The subsequent data page is located using the inner table indices or by traversing links between data pages.

Merge join process

A system for concurrent processing of queries and transactions against a shared database. The system includes multiple processors wherein a processor is available for processing queries and another processor is available for concurrently processing transactions. A query buffer is established for performing the query search while the data accessed by transactions is available in a database cache. Control logic in a database management system distinguishes between transactions and queries and initiates file-read control for reading the file containing the database. File-read control contains logic for logical sequential reading and logical non-sequential reading of the file. Control structures provide a means for control over the load that the query is allowed to place on the system.

System for optimally processing a transaction and a query to the same

Growing community of inventors

San Jose, CA, United States of America

Lee-Chin Hsu Liu