[EMAIL PROTECTED] wrote: > > I am looking at TCP4s Flow Control mechanism. I have found > Additive > Increase/multiplicative decrease and slow start explanation. > > What I am looking for is how TCP/IP on our servers behavior > when congestion > occurs related with the window size. > > Any Thoughts?
I have some thoughts, though not real answers. :-) Slow start has been implemented by most TCP implementations in most operating systems for quite a few years now. You can easily see it in action if you use a protocol analyzer on your system while doing FTP, for example. Watch the start of the data session. Even though the side receiving data will advertise a window of 8,192 bytes, for example, the other side will only send a couple segments, wait for an ACK to make sure there's no congestion, and then speed it up and send about 5 segments at a time, filling the recipient's receive window. If it's Windows, you'll see the recipient open and close its window in very small increments sometimes, for example go from 8,192 to 8,000 and back again. I always thought that was just brain-dead programming, but maybe there's a reason for it. Regarding ddditive increase and multiplicative decrease, I've never heard of any operating system doing that. After a quick Google, I get the impression those may be research projects? Can you tell us more about them? Regardless, if you know how they work, your best bet is to study traffic with an analyzer and see if you see evidence of them. In the case of Windows, I doubt you'll find any documentation that says they are used. In the case of open-source Unix TCP implementations, you might have more luck finding documentation on protocol behavior for specific implementations. Priscilla > > Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=63927&t=63904 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
