Data Management - The CIO's Guide to Oracle Products and Solutions - page 206

Guideline 12: The logical design of application systems and data-
bases should be highly partitioned. These partitions must have
logical boundaries established, and the logical boundaries must
not be violated.

1. Applications will need to be divided into coded entities (e.g., pre-

sentation, process, and data access).

2. For databases, there will be a need to develop competency in par-

titioning horizontally and vertically; this will result in more but
simpler tables and views. Design reviews must ensure that logi-
cal boundaries are kept intact. Data management responsibilities
will be increased for DBAs.

3. An increase in the analytical skills of project analysts will be
required to determine when partitioning takes place based on
expected data and/or access requirements.

Guideline 13: Online transaction processing (OLTP) should be sepa-
rated from data warehouse and other end-user computing and
Internet access.

1. Data marts represent a type of configuration standard for physi-
cal partitioning.

2. Data warehousing and data marts will need to become core com-

petencies of IT.

3. Business and IT will need to agree on the purpose and objective
of the data warehouses.

4. Data redundancy will be necessary.

5. Data marts will not reflect the most current data.

6. It will not always be necessary or even desirable to physi-
cally partition data, such as when there is a low scalability
requirement.

Guideline 14: IT solutions should use industry-proven, mainstream
technologies.

1. Criteria for vendor selection and performance measurement will

need to be established.

2. Criteria to identify the weak vendors and poor technology solu-
tions will need to be established.