What does "streaming-off-disk" mean?

Question

I am taking a course offered by www.statistics.com—Advanced Analytics and Machine Learning with Hadoop—and I reading through course material. I stumbled upon the following statement (emphasis mine):

…because MapReduce is designed to run in a batch-oriented, streaming-off-disk fashion, this presents a number of limitations when performing data analysis that entails iterative processing (including many machine-learning algorithms) or interactive data mining that requires responsive feedback.

I Googled the term “streaming off disk” and wasn’t satisfied with the first page results.

Can someone explain what it means for a programming model (e.g., MapReduce) to be “designed to run in a … streaming-off-disk fashion“ and why this fact is noteworthy (given the context above)?

Answer 1

'Disk Streaming' is the process where files are kept open while you perform multiple read or write operations within a loop. Disk Streaming is used to save memory by reducing the number of times a function interacts with the OS to file opening and closing.

For more references , please read this article "http://zone.ni.com/reference/en-XX/help/371361L-01/lvconcepts/basics_disk_streaming/".