Should linked data sets form part of your data strategy?

CIOs at a Computing event agreed that linking data can add value, but can also introduce weaknesses

How data sets are linked is more valuable than their size, CIOs agreed one of Computing's recent IT Leaders' Club, sponsored by Informatica - but the practice also introduces a single point of failure to the IT environment.

IT decision makers from a variety of different verticals said that data-sharing is key to success, but internal attitudes can mean that reality differs wildly. Even sharing information for the common good can be a difficult trait to instil:

"We have a data-sharing group to encourage people to share data internally," said the CIO of a research organisation, "and they're all terrified about what someone else is going to do: how they're going to use the data, will they get an answer before me, will they make me look stupid? I think that that same worry is also true in industry… The barrier isn't so much governance, it's the fear that someone might be smarter than you, and make better use of your data than you can."

An IT leader at a financial institution was also facing an issue with adopting a more cohesive data strategy, but said that no-one within the organisation had given the idea of collated data much thought:

"There are about 99 open databases of [international institutions'] data [that we're working with] that no-one has really thought about, or how it could be leveraged better to an advantage… Rather than think of it as data, think of it as knowledge; this is knowledge that has been accumulated."

Linking data sets means faster and easier access to information for everyone who needs it, but as well as internal pushback - or a lack of awareness of the benefits - there are other factors that stand in the way. These include both commercial and technical reasons: sometimes a data set has too much value to provide it to others for free, or the sets might be in a variety of different formats - especially if they are the result of acquisitions.

"It's not about the data," said one attendee, "it's about having the right data and making the connections between it. If you have data for the sake of it it's not much use, but what decisions does it enable you to make? What is the purpose of a data strategy? Perhaps it's to enable a decision."

The leader of IT at a large telecoms firm questioned the security of linked databases. The company wants to have "a massive database, because the architects generally think ‘I need five systems to be one system'... It's a natural reaction of an enterprise architect to link data."

He added, "I'm starting to think that unless there's a real need for them to be one system, I'd rather they weren't - because at least then the blast radius [of an attack] is only one-fifth of the size."