[dba-SQLServer] Streams

Please excuse the double post, but I'm unsure how to attack this. Also,
please forgive my ignorance on this topic.

As I understand it, a stream is essentially a file in memory rather than on
disk (although I suppose that it could overflow available ram and end up in
virtual memory, on disk).

Although I have been doing some reading on the subject, I don't think that I
grasp the essentials yet. Imagining a file in memory that is hooked to via a
recordset, how does one peel off the records from the stream and write them
to disk? I guess the important question here is, how do I peel the incoming
records off, as in a FIFO stack operation? A secondary and unimportant (for
work purposes) question is, how is a stream such as the input from BBC or
CBC radio structured? Is the whole stream one giant record or is it somehow
structured into many records?

In the case at hand, the stream of interest is a Bloomberg stock prices
stream.

All these questions arise due to the fact that our star c#/c++ programmer
had a stroke two days ago, and until we hire someone I have to go through
his stuff and attempt to comprehend it. My job definition just got a lot
bigger.

TIA,
Arthur