0:46 - congestion control is one of top five issues in computer networking. What are the other four issues ? is there a youtube lecture on that top five? Many thanks for the videos. it explains in a simple nice way a very complicated issues
It could be things like improving response times with caching, security, better utilization of the bandwidth (routing), reliability etc. not sure in what order of importance they are though, he definitely should make a video about this.
Let's say router can process 100 packets in 100sec. If router recive 2 packet, last packet will be transmitted after 2sec, if there are 10 packets, last packet will be processed after 10 sec. Hence dekay is increasing. But if router is receiving 100+x packet in 100sec, x packets will be accumulated for the next turn. This accumulation will keep increasing after every 100 sec.
@@tarunpahuja3443 Still doesn't make sense to me. Say router receives 2 packets, it takes 2 seconds to transmit them, but both senders sends at most 1 packet every 2 seconds, so the delay should be constant at 2 seconds no? And for 2 senders, the router cannot receive more than 2 packets at the same time, so there's no way you can send 10 packets at once, instead they would be sent 2 packets at a time (every 2 seconds).
Even the prof himself said in earlier videos discussing queuing delay that packets will queue only if input link tx rate is higher than that of the output link. I can't see how queuing can happen in the ideal scenario if output link capacity is R and both senders send at less than R/2. If the prof is considering fluctuating tx rates and packet delays, then I can see how queues can build up, but I feel like that should be made clear if so.
nvm, this makes a lot more sense when you consider pipelining (the longer the pipeline, the larger the delay for later packets, which increases average delay).
on 4:56 ( ƛin = ƛout ), would it be correct to say that: ƛin is the Sender’s Transport-layer READ-from-socket rate and ƛout as the Receiver's Transport-layer WRITE-to-socket rate?
Love UA-cam recommendations videos of such a topic
0:46 - congestion control is one of top five issues in computer networking. What are the other four issues ? is there a youtube lecture on that top five?
Many thanks for the videos. it explains in a simple nice way a very complicated issues
It could be things like improving response times with caching, security, better utilization of the bandwidth (routing), reliability etc. not sure in what order of importance they are though, he definitely should make a video about this.
I know he has mentioned *reliability* as another one of the top five so far.
Please do add mouse pointer on the slides, it is really hard to follow whatever you are saying with what is on the screen.
Scenario 1, as λin approach R/2, why delay is not const? and when λin large than R/2，delay become lager and lager.
Let's say router can process 100 packets in 100sec. If router recive 2 packet, last packet will be transmitted after 2sec, if there are 10 packets, last packet will be processed after 10 sec. Hence dekay is increasing. But if router is receiving 100+x packet in 100sec, x packets will be accumulated for the next turn. This accumulation will keep increasing after every 100 sec.
@@tarunpahuja3443 Still doesn't make sense to me. Say router receives 2 packets, it takes 2 seconds to transmit them, but both senders sends at most 1 packet every 2 seconds, so the delay should be constant at 2 seconds no? And for 2 senders, the router cannot receive more than 2 packets at the same time, so there's no way you can send 10 packets at once, instead they would be sent 2 packets at a time (every 2 seconds).
Even the prof himself said in earlier videos discussing queuing delay that packets will queue only if input link tx rate is higher than that of the output link. I can't see how queuing can happen in the ideal scenario if output link capacity is R and both senders send at less than R/2. If the prof is considering fluctuating tx rates and packet delays, then I can see how queues can build up, but I feel like that should be made clear if so.
nvm, this makes a lot more sense when you consider pipelining (the longer the pipeline, the larger the delay for later packets, which increases average delay).
Awesome ❤ superb sir
I wanted to tell you that wherever I am, whatever happens, I'll always think of you
hmm
Bro s in every video ,u need to chill
Great
harikasin baba
geçtin mi dersini
I'm from india. thankyou
😅
:brain
on 4:56 ( ƛin = ƛout ), would it be correct to say that:
ƛin is the Sender’s Transport-layer READ-from-socket rate
and
ƛout as the Receiver's Transport-layer WRITE-to-socket rate?