bascialy each piece of data is sent frame by frame in complete frames,
whenever you digitise anything analogue, (sound of light) you take a digital sample
a digital smaple is a representation of that particular piece of information at that given time, and at any given time (after it's captured, that sample can be sent).
if you imagine tat the sample rate of rlow quality music is 44Khz, then that means that there are 44 thousand samples a second, each can be sent almost instantaneously as a frame of data...
it's essentially the same with video.
there are technologies behind the whole setup that I'm not entierly sure of, (I couldn't tell you the different between real streaming and winamp streaming audio for example). but the basic principals remain the same.