The only thing I don't understand is initial arbitrary value set for ssthresh. I read in rfc 5681 that it's set arbitrarily high to avoid imposing limits on the max throughput of a line. I.e. "to the size of the largest possible advertised window" the language here is confusing me. If it's the largest window how can you grow beyond it.
So is TCP Tahoe and Reno just the separation between these two algorithms where one drops the current cwnd and starts anew and the other just cuts it in half but doesnt drop the previous state completely ? Or am I getting this wrong. Because the title just says TCP. It doesnt say which one.
If you want to learn someone something you need a MORE illustrations. This was like I was on class and professor just reads the lecture from his laptop/phone/notes/books... boring right? Please, more illustrations in future content!
The problem i had was the "randomness" of the congestion window but finally i got the part that is there is more than one system sending in the network so your own congestion window needs to change according to that haha, not sure why it took so long for me
This vid is so helpful and great. BTW, is there one state called "Fast Retransmission" between Slowstart and Fast Recovery that instantly resend the lost package back to sender after receive 3 dupACK?
Wooow sir what a beautiful explanation... great work sir thank you very much sir. I just gained a mass knowledge on this topic it's so interesting....!!! Thank you sir.
What if initially cwnd = 4 and in the first round one of the 4 segments is lost, how will the diagram look like then? Do I have in RTT=1: cwnd=4, RTT=2: cwnd = 7 or 8 ?