Storm data structures - map vs separated values? -
i'm using storm parse , save data kafka. data comes in identifiers , map<string,string> of varying size. after munging end goal cassandra.
should send data 1 block of tuples or split map , send each piece separately?
a tuple should represent "unit of work" next bolt in stream. if think of map single entity gets processed single, albeit complex, object map should emitted single tuple. if want different bolts independently processing different map attributes, break map subsequently processable subsets of attributes , emit multiple tuples.
Comments
Post a Comment