|
Unfortunately, there are no guides to help in defining the hardware that is required for data warehouses. Generally we look at the following components to determine hardware.xc
1) Disks
- Data storage needs (tables, indexes, temp space, etc.)
- Data landing space (for inbound files and archives)
- Software needs (database, ETL and BI software)
- BI software space requirements
2) Memory
- Database needs (database working memory and dynamic memory)
- ETL tool requirements
- BI tool needs (report requirements)
3) CPU's
- Depends on database size and implementation approach
- Can the ETL use multiple CPUs
4) RAID
- Evaluate performance requirements versus RAID approach
- Generally we want to avoid RAID 5 due to its performance impact.
5) Growthz
- Ensure hardware is expandable
- Assume that growth will exceed expectations
I know that these factors are the basic factors, but at the end of the day, you will understand the performance requirements for your hardware and based on an extrapolation of your current hardware configurations, you should be able to judge how to utilize your hardware. The key is to buy hardware that can be expanded. You may find that what appears to be the right formula upfront, does not always serve you in the long term. So no silver bullet for hardware, but by working with the hardware team you should be able to determine a hardware spec for your environment that satisfies your initial needs and prepares you for the future.
|
