Sorting Through the Clutter
Entering into this world of Big Data headfirst, I am overwhelmed with the amount of buzz and hype surrounding the topic. The other day I read the article ‘How Big Data Became so Big’ by Steve Lohr via the NY Times website and it really set the stage for the world’s challenge of Big Data. You know something has hit the big time when Dilbert references it in passing.
Per my previous post, I do not view Big Data as a product (or as a group of products), but instead as a challenge that organizations face in their journey to analyze ALL of the data made available to them to make better decisions. Hadoop is one tool to get there – yet not the only one. Over the years we have gone from machine readable punch cards to petabytes of data stored on an array of different disk types – commodity through high performance solid state.
Great – lots of storage for data – more clutter – just like my email account. Could end up being an episode of hoarders for techos.
It’s not just the analysis of the data that is important (think a superfast data warehouse appliance cranking through queries – ala Netezza) but also the determination if the data is actually worth being stored. It is like one big garage sale. There is so much to dig through, so many items old and new – You sure as heck are not going to take it all home with you – as most of the items are garbage and not needed – they would just sit around in your house (warehouse that is) and waste premium storage – and perhaps trip you up on the way to the car (or to your analytical appliance) that you have revved and raring to go
This is where hadoop comes in handy. Hadoop sorts through your ‘digital exhaust’ (as well as any other massive load of data) and culls insight or information from it. This result can then be sent to the data warehouse for analysis – It does not have to be sent there, but in most cases I’m assuming that many folks would like to include the new insights into their analytics.
Think customer churn models, if hadoop was able to determine 1 or 2 other hidden or unknown traits of a customer segment from lets say, web click through routines (the exhaust) – The analysis would be much more accurate and theoretically save (or make) the organization money.
There are many ways that hadoop technologies can be a part of your enterprise data warehouse or big data platform – this was just one simple example that I like to use to get my head around the technology.
At the end of the day, hadoop enables analysis of Big Data problems – It might not answer them all on its own – but it is a key player (if not ‘the’ key player) in Big Data Analytics.