WP_Data Management Best Practices

W H I T E P A P E R

www.persistent.com

Section

below provides additional advice for selecting tools and database technology from a performance point

7.2.2

of view.

A best practice mentioned is to

appoint a person to the role of a metadata manager

responsible for creating and

implementing themetadata strategy. Themetadata strategy will include

—

Surveying the landscape for the locations, formats, and uses of metadata across the DW/BI System

—

Working with the data steward to educate the DW/BI team about the importance of metadata and the metadata

strategy

—

Identifying and/or definemetadata that needs to be captured andmanaged

—

Deciding on the location and the version identification for eachmetadata element

—

Creating systems to capture any business or process metadata that does not currently have a home

—

Creating programs or tools, or purchasing a tool to share and synchronizemetadata as needed

—

Designing and implementing the delivery approach for getting business metadata to the user community, and

manage themetadata andmonitor usage and compliance.

We will review this metadatamanagement issue further in sectio

below.

4.4.1

4.3.3 Technical Architecture at Development Stage

At the development stage, activities center on the dimensional model development (logical and physical) described in

sections

and

the ETL development to populate the physical model described in

, the data cleansing

5.3.3 5.3.4 5.3.5

development, section

and the performance and security aspects described in sections

and

. 6.2.4 7.2.4 7.2.8

Integration of the various metadata repositories from the off-the-shelf selected tools is addressed in section

4.4.1.6

below.

We are ignoring for themoment all aspects of BI development in this first version of the document.

4.3.4 Technical Architecture at Deployment Stage

Kimball says (chapter 13, page 544) that when a developer gives you a demo working on a developer environment

with the complete scope for the first time, you should consider that the project is only 25% done. Once it passes all unit

tests, you have completed 50%. The third 25% is when you have validated and debugged a simulated production

environment, including data quality assurance, performance and operations testing. The last 25% is delivering the

system into production, which includes usability testing, documentation and training.

This is to contrast with typical testing effort estimation as a percentage of development time, which is anywhere

between 25 to 40% in traditional application programming.