Cost Justification and Budgeting the Data Warehouse

1
Cost Justification and Budgeting the Data
Warehouse
2
The Cost of the Data Warehouse

• A data warehouse can cost a company a lot of
  money.
• From thousands to millions pending on the size of
  the warehouse or the size of the company.

3
The Cost of the Data Warehouse

• The cost of the data warehouse can be extended
  from the amount of hardware and software to the
  programmers and the cost of running the
  environment.

4
Justifying the Cost of the Data Warehouse

• Justifying the cost of the data warehouse can be
  complicated for these reasons
• The benefits are hard to determine due to the
  fact we cant see the benefits until the
  warehouse is completed
• From a business stand point a data warehouse
  seems to be a good because it may produce revenue
  or save expenses, but the only way to see if a
  warehouse is worth we need to know if it saves
  time.

5
Justifying the Cost of the Data Warehouse

• From a business stand point a data warehouse
  seems to be a good because it may produce revenue
  or save expenses, but the only way to see if a
  warehouse is worth we need to know if it saves
  time.

6
Budgeting the Data Warehouse

• The reason for budgeting a data warehouse is to
  help determine what sections of the environment
  are costing the company the most money.
• A budget for a data warehouse can also give an
  idea of what the recurring expenses are.

7
Things to be determined in the Budget

• What hardware is being used the most?
• Disk storage, processing cost, network cost
• What software is being used the most?
• DBMS, access tools, metadata, integration of the
  warehouse

8
Things to be determined in the Budget

• Another deterring factor for a budget is whether
  the warehouse is centralized or distributed.
• The reason for understanding what type of
  warehouse the company has is because some costs
  for a centralized warehouse maybe less than some
  costs for a distributed warehouse.

9
Davids Question

• Some may ask why is cost justification and
  budgeting a data warehouse so important?

10
Real Time Data Warehousing
11
What is Real Time Data Warehousing?

• Updates to the data warehouse are instantaneous
• A tongue and cheek definition of real-time is
  any data delivery that is too fast for the
  current extract, transform, load system

12
Methods of Real Time Data Warehousing

• 3NF-centric approach
• Star schema
• The trickle and flip method
• Table partitioning
• Real-time partitioning

13
3NF-Centric Approach

• Focuses on the methods of operational systems
• Designs do not support query and analysis
• Upstream data repository for a series of star
  schemas, online analytic processing (OLAP) cubes,
  and other structures that are directly queried by
  analysts
• The system just adds to overhead

14
Star Schema

• Provides a one-stop load-and-query approach
• Heavy indexing and denormalization
• Warehouse data must be on the same level as the
  transactional system

15
The Trickle and Flip Method

• A copy is made of the fact table and given a name
  that isn't in the catalog of any query tool
• New records are added to the copy
• At a time interval, the copy is copied and
  renamed as the fact table for access in the
  warehouse.
• The data continues to flow into the copy

16
Table Partitioning

• The fact table is partitioned on a time key
• Data is moved into a new partition held separate
  from the queriable warehouse
• At the desired interval the partition is made
  available to the warehouse.

17
Real-Time Partitioning

• Not literal table partitioning
• Set-up of special table subjects within rules for
  updating and querying the data warehouse
• Dimension tables must also be updated in real-time

18
Matts Question

• What are the five methods to developing a
  real-time data warehouse?