In the real world, if we want to resolve a big data problem, then we have to think the solution in map reduce pattern. So we have to identify the key, value and sorting everything. Some problems can’t be run parallel it may requires some sequential flows. At that time, the Cascading API will be in handy
Please be aware that under the hood, the cascading program would be converted into Map reduce programs
Source ==> Pipe==> Sink
Source, Sink could be anything, HFS, LFS, HBase
Pipe is set of functions and operations.
It can be compared with Hive/Pig. But these tools are mainly used for Adhoc request. For doing extensive analysis, we should use Cascading API and Hadoop MapReduce