Calculating the analytics equation
Can your customer afford big data at the price of a relational database, asks Matthew Napleton
The promise of big data is timely access to business insight; the reality is a massive dent in the CFO's budget.
For any CFO concerned about throwing good money after bad on big data deployments that are simply failing to deliver the information the business needs, there are three questions.
Why does the business need to collect this data?
How is it going to provide value?
If the way IT has attempted to manage big data to date has not succeeded, isn't there a better way?
Trying to solve this essentially 21st century problem with outdated and inappropriate 20th century technology is not only futile but unnecessary, given the latest generation of database technologies.
If the CIO is not prepared to face up to the limitations of traditional relational database technology, isn't it time for the CFO to push for radical change?
The situation is set to get worse. Most CIOs have yet to wrestle back some control over structured data to drive operational improvements, and Forrester Research has claimed that nearly half of all big data deployments are just for marketing purposes.
How can that be justified? How can an improvement in sentiment analysis from social media, for example, deliver bottom-line benefits to a retailer unable to gain the insight required to improve day-to-day logistics and stock control? Once again, the CFO is being asked to pile even more investment into the big data project.
Organisations have been struggling with growing data volumes for years, despite continued advances in often-expensive technology.
Organisations always seem to have too much data to analyse and not enough computing power to do so in a reasonable time.
But organisations are trying to collect the data from too many sources, often without good cause, and they are attempting to analyse that data with the tools they have used for decades.
The result is an entry-level investment point far beyond the budget of all but the largest organisations and a technology model that is simply not fit for purpose.
The relational database (RDBMS) has a place and a purpose, but it is not a tool that was ever designed for the data volumes it is now being asked to handle. Despite upgrades, add-ons and so on, when it comes to big data the RDBMS may have had its day.
Newer database technologies have been specifically designed for the new era of data management, to manage and compress vast data volumes that can be mined for insight.
It may require just a server, not an entire datacentre, to run. Innovative data compression and pattern matching requires less infrastructure and is therefore cheaper.
There is nothing wrong with the concept of turning data into valuable information. The problem has been the way the IT industry has attempted to achieve this goal.
At best, the big data explosion has become a distraction from the need for businesses to use information to drive operational improvements. At worst, CIOs have been coerced into spending big to capture unstructured social data – despite a clear lack of success in managing structured operational data – for fear of falling behind.
It is time to accept that attempting to manage fast-escalating data volumes with traditional RDBMS does not work.
Matthew Napleton is marketing director at Zizo