This article originally appeared on the BeyeNETWORK.
Recently, there has been a flurry of information about data warehouse appliances and their benefits. The information presented in articles and white papers has repeatedly shown the benefits of this technology, including:
- Lower cost – initial and ongoing
- Higher sustained performance
- Ease of use
Putting all of this aside, the hard question to ask is this: Why do we need data warehouse appliances? Before we answer this, let us look at the overall cost of a data warehouse. There are several critical components that can be broadly classified into the following:
- Initial cost: server hardware, licenses, development, deployment
- Ongoing cost: maintenance, upgrades, hardware and storage additions
- Data cost: initial data volume, incremental transactional volume, mergers and acquisitions
- Business cost: data requirements – granularity, near real time, historical data retention
(Author's note: Business intelligence tools and their costs are not included in this discussion.)
There are costs that are necessary and cannot be controlled, such as:
- Initial cost of hardware
- Initial cost of development and deployment
- Initial cost of software licenses
- Ongoing cost for upgrades and maintenance
- Ongoing development cost
And there are costs that can be controlled, such as:
- Ongoing storage expansion
- Ongoing server expansion
One might be tempted to ask: If we conclude that just by adding a data warehouse appliance, we are magically saving dollars and reducing cost, is this justification enough to adapt to a new technology and bring it into the data warehouse ecosystem?
Figure 1 is a sample of costs in the data warehouse; the comparison is spend per dollar over a period of time.
Figure 1: Initial Cost of Data Warehouse
From this graph, you can see that we are constantly in the process of spending money in the first year of the data warehouse build, which is normal for any new IT initiative. But as we progress through the timeline, we see the consistent increase in spend for storage and services. We can argue that as the data warehouse maturity happens, it is natural to spend more on storage and services. This should be an anticipated expenditure.
Figure 2: Data Warehouse Maturity – User Growth
Figure 3: Data Warehouse Maturity – Data Growth
Figure 4: Data Warehouse Maturity – Response Time
As shown in Figures 2, 3 and 4, we can see that with increase of adoption, performance and availability become issues. To mitigate these issues on the current implementation, there is the requirement for more investment into the hardware, storage and software areas. This provides interim relief and is not a long-term solution. This is exactly where the spending cycle starts.
But, wait a minute! Where is the cost coming from? Disk is cheap and processor is cheap. It is only the need for additional license fees for software that is causing the need for additional spending, correct? Actually, not quite correct, and that will be covered in the next article in this series.
Part 2 of this article will continue our look at why there is a need for data warehouse appliances and will examine data warehouse costs from an infrastructure perspective.