X
Business

The true cost of analysis

When developing a data warehouse, you effectively face three choices: expensive, ridiculously expensive, or ludicrously expensive.
Written by Angus Kidman, Contributor

When developing a data warehouse, you effectively face three choices: expensive, ridiculously expensive, or ludicrously expensive.

However, that doesn't mean that you can't apply some semblance of fiscal logic to the installation decision.

At Sybase's TechWave conference in Las Vegas this week, IBM's P Series general manager Ross Mauri offered an undoubtedly biased overview of the kinds of costs that are involved. (The fact that IBM, a rival of Sybase's in the pure database industry for years, was willing to do some serious flag-waving at a Sybase event suggests that the late '90s concept of co-opetition is very much alive and well, by the by.)

Mauri weighed up the relative costs involved in building a system that would allow 50 users to do real-time, ad hoc analysis against a five terabyte database.

By his estimate, a Sybase IQ system running (naturally) on P Series hardware would cost around US$270,000, while a BI appliance would cost around US$700,000, and a Teradata Clique system a cool US$1.4 million.

The point Mauri wanted to make was that technology decisions can significantly influence installation cost. But what struck me was how, even on the cheapest option, those ad-hoc analyses were going to have to produce some serious and tangible business results.

On a per-analyst basis, the IQ option weighs in at US$5,400 -- and that doesn't include the actual salary paid to the number cruncher, or the ongoing IT support costs for running the system. To justify that, you're going to have to be doing more than just ensuring regulatory compliance or spotting occasional cost problems.

Conversely, if every analyst can produce double their own salary in measurable savings through intelligent use of data, then the US$1.4 million upper level price tag might look cheap.

As ever, you need to look at lifetime costs (and likely lifetime outcomes) before you can make an intelligent decision -- and at these rates, you can't afford a lot of analysis to get there.

Angus Kidman travelled to Las Vegas as a guest of Sybase 365.

Editorial standards