Hardware

Data warehouse queries often require scanning millions of records so, as far as hardware is concerned, you need to maximize the amount of memory in your RDBMS server, and use the fastest hard drives your money can buy.

Why should you maximize the amount of memory in your server? The key here is that if the RDBMS can load the most frequently used tables into memory, querying a table stored in memory will be much faster than querying a table stored in disk.

Other issues that I'll write about later:

• CPU speed
• Number of CPUs
• RAID levels
  
• MPP vs SMP architecture
• The Netezza approach