International Business Machines Corporation

David Glenn Carlson

Kevin James Kathmann

Robert J Bestgen

Steven Virgil Hoeschen

Shantan Kethireddy

Jeffrey S McAllister

Robert Victor Downer

Nelson Ramirez

Kevin Curtis Griffin

William Anton Brown

Lee Anton Sendelbach

Michael Joseph Corrigan

Ulrich Thiemann

Andrew Peter Passe

Timothy John Mullins

Wei Hu

David Lee Hermsmeier

Travis Michael Drucker

Paul Richard Chmielewski

Thomas Owen McKinley

Glen Warren Nelson

Thomas Paul Giordano

Thomas E Miller

Lawrence Kenneth Nash

Tariq Mahmood Choudhry

Thomas Edward Miller

Embodiments are provided for deferring allocation of storage for segments. The method includes receiving, at a first moment in time, a request to allocate a first segment within a memory. Information corresponding to the first segment is stored in a first entry in a hash table, and memory is allocated for the first segment. Information about the first segment is returned to an entity that requested the allocation of the first segment, where the first segment is not allocated auxiliary storage until predefined criteria are satisfied. At a second moment in time, upon receiving an indication to deallocate the first segment, memory corresponding to the first segment is deallocated, where the first segment is never allocated auxiliary storage and the first entry in the hash table is deleted.

Deferred method of allocating disk space for lightning segments

Techniques for recycling segment pages while preserving integrity of memory addressing. Upon a request to delete a system object, an initial segment identifier is de-assigned from a segment of the system object, and each segment page of the system object is remapped to a new virtual address range. The segment is stored as a reusable segment in a reusable segment store. Upon a request to access the system object, an exception is raised that the system object does not exist. Upon a request to create a new system object, the reusable segment store is searched to identify the reusable segment as a match. The reusable segment is assigned as a segment of the new system object.

Recycling segment pages while preserving integrity of memory addressing

A method, apparatus, and program product stream data between a host element and a target element in a parallel computing system. The method comprises partitioning data used to perform a task into a plurality of data streams based at least in part upon data associated with the task and partitioning a data stream into a plurality of data stream views based at least in part on a memory size associated with a target element configured to process the data stream. The method further comprises sequentially communicating at least a portion of the data stream views to the target element via a data streaming infrastructure implemented on at least a host element and the target element, wherein the portion of the data streaming infrastructure on the host element transfers the portion of the data stream views to the portion of the data streaming infrastructure on the target element.

Data streaming infrastructure for remote execution in a constrained environment

A method, apparatus and program product are provided for performing a query of a database. A database query is received, which includes first and second operations respectively configured to generate first and second disjoint results sets from a dataspace. The database query is analyzed to identify a set of attributes from the dataspace that are used by at least one of the first and second operations in the database query. During execution of the database query, a plurality of records from the dataspace is iteratively processed in a single pass, including, for each of the plurality of records, processing such record by retrieving the plurality of attributes for such record from the dataspace and performing each of the first and second operations on the record using the retrieved attributes for such record to build the first and second disjoint results sets.

Multiple result sets generated from single pass through a dataspace

Nodes in a query execution data structure have monitor and dump methods defined. The monitor method enables the collection of information from the node. The dump method outputs the monitored information from each node. A monitor and debug mechanism of the preferred embodiments includes a graphical user interface that allows a user to graphically examine a query execution tree, to enable monitoring of the nodes on a node-by-node basis, and to view information dumped from the query execution data structure as the query is executed or after the query is executed. The result is a powerful tool that allows efficiently monitoring and debugging a query implemented in an object oriented query execution data structure.

Monitoring and debugging query execution objects

Techniques are disclosed for converting data into a format tailored for efficient multidimensional fast Fourier transforms (FFTS) on single instruction, multiple data (SIMD) multi-core processor architectures. The technique includes converting data from a multidimensional array stored in a conventional row-major order into SIMD format. Converted data in SIMD format consists of a sequence of blocks, where each block interleaves s rows such that SIMD vector processors may operate on s rows simultaneously. As a result, the converted data in SIMD format enables smaller-sized 1D FFTs to be optimized in SIMD multi-core processor architectures.

Processing array data on SIMD multi-core processor architectures

A stream data node receives real-time streamed data from one or more input devices, dynamically filters the streamed data to reduce the streamed data, and delivers the reduced data when requested. By providing real-time filtering of the data, the amount of data that must be stored in a database may be substantially reduced. The stream data node can perform aggregation functions, group functions, and select functions, thereby also significantly reducing the amount of data that must be stored in a database. The stream data node may also be part of a query execution data structure, where it delivers its data when requested by another node in the query execution data structure.

Real-time mining and reduction of streamed data

A method, apparatus, and program product retrieve data for a task utilizing demand-based direct memory access ('DMA') requests. The method comprises, prior to the execution thereof, analyzing a first portion of a task to determine whether data required for execution thereby is stored in a local memory, and, in response to determining that the data required for execution by the first portion of the task is not stored in the local memory, proactively issuing a first DMA request for the data required for execution by the first portion of the task. The method further comprises, in response to determining that the first DMA request is not complete, determining whether to proactively analyze a second portion of the task prior to the execution thereof for a determination whether data required for execution thereby is stored in the local memory.

Demand-based DMA issuance for execution overlap

A method, apparatus and program product dynamically activate and/or deactivate buffers during execution of a database query. The dynamic deactivation and activation is based on the monitoring of information associated with the consumption of data by a plurality of threads during execution of an access plan for the database query. By doing so, the selection of a less optimal buffer position during optimization may be replaced with a more optimal buffer position during runtime, often resulting in improved query performance and parallelism.

Database query processing with dynamically activated buffers

A process combines multiple grouping sets into single rollup sets with depth lists defining the levels of grouping that must be performed. Grouping sets are identified that are contained within other sets and combined into single rollups with depth lists. Cube aggregation conversion to rollup aggregation is provided for optimizing database query processing. Natural sets of rollup hierarchies within a cube are recognized and the cube is converted into those rollup hierarchies. Once converted, the rollup aggregation is performed to significantly reduce required processing.

Implementing aggregation combination using aggregate depth lists and cube aggregation conversion to rollup aggregation for optimizing query processing

Methods, systems and articles of manufacture of processing a multi-state attribute field query. One embodiment provides a method of processing a multi-state attribute field query, comprising instantiating a plurality of live objects; performing, by each live object, an attribute operation, wherein at least one attribute operation is a comparison operation associated with attributes of the query. The plurality of live objects access a global status storage area only when an operand of the comparison operation is null or error. In any case, execution control is passed from each live object to an adjacent live object.

Growing community of inventors

Rochester, MN, United States of America

David Glenn Carlson