I was wondering what is so 'big deal' about big data. Data Miners have been used to mine large data sets since early 2000s. Since, data miners anyway resort to stratified random sampling, how does 'big data' pose a new problem now? We know well that some algorithms scale-up well and some do not. Unless someone wants to mine data in memory - the utility of which I doubt- we do not need this new buzz word. I could be wrong. Pls enlighten me.