Business Intelligence and Data Analytics: An Overview of Data Warehousing and Real-time Online Analytical Processing (OLAP) Systems

Instructor: Greg Cermak, Axian, Inc.
Course Number: 04-CPD-0930
Dates/Times: Tuesday, Sept. 30; Networking @ 6:15 PM; Lecture @ 7:00 PM
Location: OGI School of Science & Engineering
http://www.cpd.ogi.edu/ Click headings - Public Seminar & Database Programming

Abstract:
Online analytical processing (OLAP) is typically defined as the processing and analysis of shared multidimensional data. In practice, OLAP systems analyze data drawn from large, low-transaction, high-latency relational databases, such as data warehouses. The purpose of such analysis is to aggregate and organize business information into a readily accessible, easy to use multidimensional structure. OLAP systems store some or all of this aggregated information either within tables in a relational database (also known as relational OLAP, or ROLAP, storage) or in specialized data structures in multidimensional databases (also known as multidimensional OLAP, or MOLAP, storage). OLAP queries can be answered much more quickly than similar relational queries because the aggregations and computations have already been completed and the resulting derived values are readily available from a ROLAP table or MOLAP storage.

Retrieving, analyzing, and aggregating large amounts of historical data can consume extensive time and resources. OLAP systems do not usually run against online transaction processing (OLTP) or other high-transaction, low-latency databases because the time and resources required can affect the performance of the relational database. Instead, OLAP systems typically run against data warehouses, which are updated relatively infrequently, to support the requirements of most commercial and financial analysis. Most OLAP systems rely on a "snapshot" approach, periodically retrieving and aggregating data for later presentation and analysis. Because OLAP systems typically rely on stored, derived values to answer queries, the aggregation process must also reasonably match the update latency of the underlying relational data source to avoid presenting overly "stale" data.

Products that can perform aggregations quickly enough to provide multidimensional data from low-latency data sources have challenged this traditional view of OLAP in recent years. This functionality, which is referred to as real-time OLAP, is most often used in financial or industrial scenarios where multidimensional analysis of low-latency data is crucial to the organization's business intelligence requirements.

Bio:
Greg Cermak is a Microsoft Certified Trainer and Solution Developer with more than twenty years' experience providing consulting and training in developing high-performance online transaction processing (OLTP) and online analytical processing (OLAP) systems. His specialty is Microsoft SQL Server RDBMS and Analysis Services. His interests include technology, reading, history, bicycling and robotic exploration of the solar system. He is a Solar System Ambassador for the NASA Jet Propulsion Laboratory (JPL). He is a frequent speaker at school programs, public, and industry events.