Metrics that should be developed to measure the performance of a data warehouse in the following environments: 1) health care, 2) retail and 3) pharma.

Essay by kamer06University, Master's March 2004

download word file, 6 pages 1.0

Downloaded 158 times

In most of the Database guide papers, eight performance-based metrics were used to communicate areas of concern to the data warehousing vendor. Also, the ninth one was developed as a weighted average measure using four of the eight metrics, which were deemed the most critical to the organization assessing data warehouse performance to determine how the vendor stacked up relative to data quality and systems quality.

The nine metrics that were developed to measure Solectron's, which is a major contract manufacturer in the electronic industry, data warehouse performance are;

1. Completeness

Total Complete = [((#Recordsi)/(Total #Records Expectedi) x (#Records Expectedi/Total #Records in all tables)) x 100.

This metric is supposed to serve the firm better than a simple binary yes/no in measuring the weighted average of each table in the data warehouse. It also measures the percentage completion during an Extract -Transfer- Load (ETL) and/or refresh process. This metric also, and helps identify any possible SQL problems within a particular table.

2. Connectivity

Response Time = #Users x (RTD% + TR%)

Where Run Time Duration (RTD) in minutes = Hardware/DW size

Transfer Rate (TR) = Actual Transfer Rate/Optimal Transfer Rate

This metric measures connectivity by enabling the firm to identify and isolate underlying issues

including ISP transfer rates, data sources, and hardware issues.

3. Data Integrity

Data Quality Tool Rate/Manual = % Accuracy

This metric measures the avoidance of the tedious, resource intensive method done in the past.

4. Usage Data from the Data Warehouse

Usage Rate on Data = Most frequently used/Total

This metric tracks the usage rate of tables, reports, and the overall data warehouse to identify any trends and/or problems that may occur, particularly during peak operating times.

5. On-Time delivery

This current metric is defined as: tables are on time if they are delivered by 7:00AM...