Managing sparsity in an multidimensional data structure

• US 10,169,406 B2
• Filed: 09/17/2014
• Issued: 01/01/2019
• Est. Priority Date: 07/22/2013
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Bharti, Vijay, Joseph, Jeremiah, Sivaswamy, Hemant K., Sapna, Kamalpreet, Jain, Rohit
• Primary Examiner(s): Jami, Hares
• Assistant Examiner(s): Sultana, Nargis
• Application Number:
  US14/488,517
• Publication Number:
  US 20150026116A1
• Time in Patent Office: 1,567 Days
• Field of Search: 707958, 707769
• CPC Class Codes: G06F 16/245 Query processing
• CPC Class Codes: G06F 16/283 Multi-dimensional databases or data warehouses, e.g. MOLAP or ROLAP
• CPC Class Codes: G06F 16/284 Relational databases

Abstract

Embodiments of the present invention relate to a method, program product, and system for managing the percentage of unpopulated cells in a multidimensional data structure during the servicing of multidimensional analytical queries. In an embodiment, a multidimensional data structure is stored in a memory location of a computing device, wherein the multidimensional data structure includes a dimension that is a subset of a plurality of dimensions included in a relational database. The computing device determines a frequency of utilization of a dimension to service a multidimensional query request. The computing device combines the dimension with the multidimensional data structure responsive to the frequency of utilization.

Claims

1. A computer-implemented method comprising:

determining a set of dimensions within a relational database to be stored in a cache, the set of dimensions having a historical access frequency meeting a pre-defined cardinality threshold of data-containing cells;

responsive to determining the set of dimensions, restricting an initial load to a size below a threshold size by causing a multidimensional database management system to generate the initial load using only the set of dimensions having a historical access frequency meeting a pre-defined cardinality threshold of data-containing cells;

at predetermined time intervals, retrieving log files including usage information for individual dimensions in the set of dimensions stored in a cache and in uncached dimensions stored in a relational database, the usage information corresponding to a level of usage of the individual dimensions;

at the predetermined time intervals, determining at least one relatively high usage dimension in the uncached dimensions based, at least in part, upon the usage information; and

at the predetermined time intervals, causing the multidimensional database management system to refresh, in a memory device, the initial load using only a portion of the set of dimensions and the at least one relatively high usage dimension in the uncached dimensions from the usage information in the log files, the refreshing action generating an incremental load to be stored in the cache.

2. The method of claim 1, wherein the initial load is an initial online analytics processing (OLAP) cube.

3. The method of claim 1, wherein the initial load and the incremental load are in the form of online analytical processing (OLAP) structure.

4. The method of claim 3, further comprising:

selectively providing access to the OLAP structure and to the relational database.

5. The method of claim 4 wherein the selective provision of access to the OLAP structure and to the relational database includes:

on condition that the data sought to be accessed is included exclusively in the at least one relatively high usage dimension, access is provided by accessing the OLAP structure using a Multidimensional Expressions mode query; and

on condition that the data sought to be accessed is not included exclusively in the at least one relatively high usage dimension, access is provided by accessing the relational database using an SQL (structured query language) mode query.

6. The method of claim 4, wherein the historical access frequency is based on log files including usage information that reflects how many times dimensions in the set of dimensions has been accessed since the OLAP structure was last generated.