IBM PureSystems Centre

Extending the value of IBM PureSystems

DBXten DataBlade for Data Series

Barrodale Computing Services Ltd.
System:
IBM PureFlex System
Provided by:Barrodale Computing Services Ltd.
Request Information

Version:
DBXten is a database extension that provides very efficient management of single- or multi-dimensional data series in a SQL database. A data series is any sequence of numeric or textual data, such as time series, maps, gridded data, molecular structures, etc. Now enabled for PureSystems for integrated expertise.
  • - Business Value

    Pattern Overview


    Barrodale Computing Services Ltd. (BCS) has developed a (US patented) technology that enables users to efficiently and rapidly store, retrieve and manipulate data series contained in a database management system (DBMS). The data series can be either single-dimensional or multi-dimensional, and the dimensions, while often being spatial or temporal, can be from any numeric or date-time domain.

    The DBXten technology achieves its performance gains by storing data column-wise in blocks and using multi-dimensional indexes to quickly retrieve the blocks of interest. Within each block, data is stored in a highly compressed fashion, optionally taking advantage of known limited precision in the data. However, even though the data is stored in blocks, the DBXten interface can present the illusion that the data is stored in a more natural row, or tuple, form.

    This technology has been implemented for use with the IBM® Informix Dynamic Server 9.x, 10.x, and 11.x DBMS product as a DataBlade, which is a collection of datatypes, their associated functions and operators, and access methods. To date, DBXten has also been implemented with Oracle 11g and PostgreSQL.

    Business Case

    DBXten provides diverse business opportunities in, for example, financial analysis, meteorology, GIS, geophysical prospecting, and seabed observatories, since they all involve storage and analysis of very large multi-dimensional data series (e.g., time series for stocks, 4D weather grids, molecular structures, and hydrophone array output).

    In recent years there has been an explosion in continuously generated measured data, and often such data arrives at high rates. For example, an acoustic array may include a few hundred hydrophones, each producing data at 100,000 samples per second, and a video camera might produce 640x400 samples 30 times a second (more than 7,000,000 samples per second).

    DBXten was designed to store large volumes of sequential data (i.e., data series) in a SQL database in a compact form that is easily searched, quickly accessed, and efficiently archived. DBXten can speed up insertion of rows more than 50-fold, decrease index sizes by a factor of 1000, generate these indexes more than 1000 times faster, and decrease database table sizes 10-fold.

    In contrast, ingesting high-volume data into a SQL database that is not equipped with DBXten has a number of storage trade-offs when done in either of the two conventional ways.

    Although many users still prefer to store and manipulate data series using flat files, DBXten provides an opportunity to promote the advantages of a DBMS (data integrity and consistency, independent and effective access to data by diverse users of multiple applications) since these benefits can now be achieved without loss of performance relative to flat file usage.

    Business Problem

    The business problem at hand is to facilitate the storage of large volumes of sequential data (often arriving at high rates) in a SQL database in a compact form that is easily searched, quickly accessed, and efficiently archived. Examples of sequential data include time series, N-dimensional grids (including raster images), molecular structures, hydrophone array output, GIS spatial objects, financial data, and meteorological measurements.

  • -Technical Details

    The DBXten DataBlade provides facilities for (i) rapidly ingesting and indexing data series, (ii) in response to a multi-dimensional query, identifying and retrieving just those series elements that satisfy the query, and (iii) casting data seamlessly between compressed blocks and sets of conventional database table rows.

  • - More Information

    Provider Information


    Name: Ian Barrodale

    Email: BCSinfo@barrodale.com

    Phone: 1 250 412 7428

PureSystems patterns are part of a broad portfolio of PureSystems solutions for business, cloud, and infrastructure applications.

See more solutions

Related tagged content

Industry

You must  login with your IBM ID to be able to bookmark. Select "Login" to bookmark or "Close" to return.