In general, you use the phases of a M/R like so: 1)

getInputData

retrieves all necessary data from the database (usually via search), combines all relevant data into a single data structure, and returns that data structure 2)

map

[optional] receives the data and organizes it into groups (e.g. takes the Invoices returned by

getInputData

and groups them all by Customer) 3)

reduce

receives the data and performs the actual data processing (e.g. storing the data in an Item field in your case) 4)

summarize

[optional] contains statistics for all previous phases and can be used to send out summary reports or notifications that the processing is complete

Not sure I completely agree Eric, but very nice. A few things ... reduce is also [optional]. I'd characterize map as "An opportunity to process each result individually, with an opportunity to group with other results based on a key and send to the 'reduce' stage"

You must use one of

map

or `reduce`; personally I always use

reduce

, hence saying

map

is optional and

reduce

is not`. Yes, technically, you can use either

map

or

reduce