title says most of it...

I'm studying for CCNA and refamiliarizing myself with the OSI model.

Never had to know much of the lower level stuff on it when I went through my MCSE track. I'm having trouble getting my head around the concept of Windowing in a connection-oriented session in Layer 4.

Namely: the book says that:
<snip>
- <data throughput> would be slow if the transmitting machine had to wait for an ACK after sending each segment. Because there's time available AFTER the sender transmits the data segment and BEFORE it finishes processing ACKs from the receiving machine, the sender uses the break as an opportunity to transmit more data.
</snip>

How in the world does this not cause the entire session to go T.U? Wouldn't this be a non-connection-oriented session inside of a connection-oriented session???!

Anyone got a different view on this?

ok, so I'll answer my own question since I now understand it....

basically, windowing allows the packets that are sent during that time period <see snip above> to be reassembled later and then when the CRC runs it'll pick up on any missing packets and THEN request them to be resent.

Apparently, it's a sort of delayed-confirmation connection-oriented session.

This takes advantage of the statistical improbability of packets being lost on a well-designed network. If you're running a huge collision domain you probably want to go with a small windowing structure.

Of course, if you're running a huge collision domain such that packets get lost that often, you're either:

1) about to get rid of all those damn hubs and go switched

or

2) going to be looking for a job soon.