Tuesday 21 August 2012

Improving the Design

  • Table structures are cleaned up to eliminate troublesome initial partial and transitive dependencies.
  • Normalization cannot, by itself be relied on to make good designs.
  • It is valuable because its use helps eliminate data redundancies
  • Issues to address to produce a good normalized set of tables:
    • Evaluate Primary Key Assignment
      • Create JOB_CODE and JOB table to reduce data entry errors
      • E.g. Database designer in not mistakenly written as DB Designer
    • Evaluate Naming Conventions
      • Keep consistent with table name
      • So CHG_HOUR-> JOB_CHG_HOUR
    • Refine Attribute Atomicity
      • An atomic attribute is one that cannot be subdivided and is said to display atomicity
      • EMP_NAME becomes EMP_FNAME, EMP_LNAME
    • Identify New Attributes that might be useful
      • For example add EMP_HIREDATE or data about address etc. Identify New Relationship.
      • Project and Employee tables have new relationships that may allow you to identify data about the project manager.
    • Refine Primary Keys as Required for Data Granularity
      • Granularity is the level of detail represented by the data. Data stored at their lowest level of granularity are said to be "atomic data".
      • Use ASSIGN_NUM to get greater accuracy and granularity to avoid entity integrity violations.
      • (e.g. what if employee worked on project at different times? then there is no longer a unique primary key if composite key PROJ_NUM and EMP_NUM is used.)
    • Maintain Historical Accuracy
      • Identify past and current charge rates using ASSIGN_CHG_HOUR instead of JOB_CHG_HOUR
      • This ensures you have accurate record of what happened at the time of the project
    • Evaluate Using Derive Attributes
      • In large database, you might want to store derived attributes to reduce time to produce reports.

No comments:

Post a Comment