Selecting The Right DW Platform

An enterprise data warehouse relies on the server platform to provide complementary performance.

Selecting the right data warehouse platform is a key part of infrastructure success. Adopting a fact-based approach is essential in comparing one DW platform from another.

1. Identify business requirements - current and future
2. Quantify technical and scalability requirements in all dimensions
3. Compare platform options directly to the requirements
4. RFP process - gain proof of meeting business requirments
5. Use Case Vendor Presentations - explanation of how architecture will handle specifi queries and solve specific scalability problems
6. Demonstration - performance and scalability with full-scale benchmark using YOUR queries and data [where possible]
7. Gain product references from current users
8. Gain vendor commitment [with penalties] to meeting specific database performance goals

Determining Requirements

To quantify the size and structure of data, determine:

• Where data is sourced from
• What data structures look like
• Data relationships
• Data update cycles
• Data growth rates

The understand the workload performance, identify:

• Business questions asked of the data - query complexity, query volume, data tables to be accessed
• Business users asking the questions - how many, access modes
• How often questions need answers - data update requirements [data latency]
• How quick the answer is needed - latency, analytics performance speed
• Business impact if data is not available, or system is slow or unavailable.

DW Platform Solutions

Leading Enterprise DW grade platforms include:

• Microsoft SQL Server 2008 - significant performance increases over 2005 version.
• Teradata's 5450 Server - features a boost of 18 per cent in the price/performance ratio over previous models and more powerful utilisation of Teradata's market-leading analytical software.
• IBM
• HP - BI and data integration. Neoview product combines HP hardware with a NonStop-inspired massively parallel database.
  

Also See Evolution of Database - Analytics Platform Models

Back To Top