By Chris Brown, big data analytics expert, OCF.
If you associate big data exclusively with large volumes of data, you wouldn't be alone. A 2012 survey by Harris Interactive of 154 companies found that nearly 70 per cent used a volume-based definition for big data. 25 per cent defined big data as 'massive growth of transaction data', 24 per cent think it refers to new technologies for managing massive data, and 19 per cent defined it as the 'requirement to store and archive data for regulatory compliance'.
Much like 'cloud', big data is a great headline grabber, but it gives off the wrong connotation – volume, volume and more volume! It's high time we started to use more informative descriptions if manufacturing firms are going accept it as a serious tool to support their operations. It would be understandable if manufacturers – particularly the smaller and mid-size firms – thought, what's big data got to do with me?
There has been, for many years, more defining words for big data. First introduced way back in 2001 and generally attributed to former Meta Group, now Gartner analyst Doug Laney, who defined big data as management challenges thrown up in the three areas of Volume, Variety and Velocity of data [the three Vs].
Of course since 2001, plenty of people have stepped forward and made additional definitions such as Value – the value you can get from big data and also Veracity – the reliability and usefulness of the data. I believe some industry commentators have even escalated big data up to eleven Vs now!
I certainly think greater use of these types of definitions is a step in the right direction, to help organisations begin to understand big data better. The effect would be similar to exchanging the term 'cloud' for IaaS, PaaS, SaaS etc – these terms are now widely used to engender a more practical view of cloud services.
However, perhaps more importantly, I'm of the belief that most of these current views are actually out of touch with modern UK manufacturing businesses. The concept of Volume/Velocity/Veracity etc. simply doesn't work every time and the volume-only definition definitely doesn't attract interest!
A September 2013 survey by supplier Steria suggested that only 8 per cent of respondents felt that 'scaling options' relating to rapidly growing data volumes represents a serious problem. The majority of companies simply do not generate, want or have access to petabytes of data (petabytes seem to be the standard units of measure for big data). Who has that much data in-memory or stored on a disk? Big data volume really only applies to a "few" select organisations, the giants such as NASA, Tesco and Wal-Mart.
Sylvan H. Morley III, Director of Information Lifecycle Governance for IBM Worldwide tends to agree:
"I find a great deal of people who hear and can regurgitate the words "Big data" but if you listen to what they are saying what they mean is "a lot of data" rather than the "big value of data". I think this misses the point, if you look at the core of what "Big data" is really about; it's about extracting value from data, the big ideas that data can provide or the big insights that data can bring that we weren't getting before. So for me "big data" is about using only your valuable data; not just managing crushing amounts of data, because at the end of the day volume doesn't really provide value by itself."
In the case of the second original V [Velocity] my issue is much the same, not every business requires the capability to stream petabytes of data in real-time. It's a requirement in 5-10 per cent of organisations we come across [and not forgetting our heritage is working in the high performance computing industry where crunching data is every day life].
Big data variety
Where should manufacturing firms be looking? Well, most companies, even the smallest, do now have a variety of data – email, accounting systems, engineering notes, images, social media feeds, audio, video, etc. which are not aggregated and put to use. It's 'big' in terms of its complexity and presents a real opportunity for manufacturing firms to build models around their company and products. Collection and analysis of big data variety can help manufacturers to calculate product warranties and predict potential product maintenance issues and requirements to stock spare parts. This prediction could even go as far as issuing product recalls or announcing planned upgrades at specific times. These firms can use big data variety to improve their customer service function.
Equally, manufacturing firms can take data feeds from their own production line to create models to predict when plant machinery might breakdown – this helps to reduce costly stoppages, which can run into millions of pounds for the largest firms. Some of the largest firms in the UK are already pioneering this.
As suppliers and speaking to the wider industry, we need to take big data and break it down, so it's more widely understood. Second, we really need to start pushing the benefits of big data Variety aspect. This will make big data more attractive to the broader base of manufacturing organisations.
For business owners out there who think big data doesn't apply, ignore the headline-grabbing name and start to think about all of the [multiple gigabytes or terabytes of] data sitting idle in your organisation. Think of the cost of lost production. Using big data analytics is no longer a concept or too expensive, it's possible now. Pull this data together – transform and analyse to deliver improvements to your business.