Basically I started understanding about the architecture of Fatwire analytics, where I need to know how the BIG data will be processed. Big means really Very Big volumes of data. Hadoop framework helps to process the data by creating Namenodes and Data nodes. The name nodes will be maintaining the metadata this could be sorted and indexed to process newly available data into local disk. further this can be handled by tasktrackers and jobtrackers they are like agents. JobTracker will assign the task to perform. this is design in Java to work on PETA bytes of data with MapReduce design pattern.
Where do we use Hadoop?
For any web site, The social marketing guy need to know 'What is the aricle most viewed?' or 'What page popular?' What page got how many impressions etc.., To do this there will be a process that senses that page visitors and their basic information can be collected by a sensor. when there is a new visitor need to collect where visitor is from? and which browser or device using etc.
When webmaster tool interacts with Analytics basing on Big Data collected from Content delivery systems. It is provided to content editor what to give the effective keywords required got most views/ customers.