Development and Management of Data Warehouse projects

dev_and_pm_data_warehouse.txt

Many DW projects fail and are abandoned midstream
Fail due to inadequate project management
Evaluate if really needed, evaluate opportunity and viability

Compare cost of failure vs project not started at all
Bottom up vs Top down
  Enterprise-wide centralized data warehouse
  individual data marts first
    faster cheaper, more limited benefits
100% in house programming (not a good idea)
100% Ready to go - unrealistic
  vendor tools available for all data warehouse functions and components
  right mix between vendor tools and in-house programming
  
When buying
  Single or Multiple Vendors or Best-of-breed

Driven based on requirements, not technology
 What each group does?
 What systems are used?
 KPIs?
 Customers?
 Data is tracked?
 Business products?
 Products and service categories?
 Locations?
 P&L measurement levels?
 Queries and reports used for strategic information?
 
Audit source systems, review its architecture, identify data sources and relationships between them
Determine how to extract sources

Justifications:
Costs are substantial, weigh costs against tangible and intangible benefits
Typical costs: Admin 10% / Hardware 31% / Software with DBMS 24% / Staff and System Implementation 35%

DW will reduce costs (Calculate costs of generating strategic information from current system, compare to DW estimated cost
DW will improve business (Calculate the business value of project in terms of market indicator improvements
DW will be more cost effective overall: Perform analysis of costs and estimated benefits then calculate return on investment ROI