2,415 views
3 votes
3 votes
Case 1: If congestion window size=100MSS, rwnd=16MSS and RTT=1ms. Then how many time it will take to send full window?

case2: If congestion window size=16MSS, rwnd=100MSS and RTT=1ms. Then how many time it will take to send full window?

How to decide intial threshold value when there is "congestion" or "no congestion" in the underlying network?

1 Answer

Best answer
4 votes
4 votes

Sender window = minimum ( Congestion window, Receiver window)

  • If RW > CW, then sender window = CW and congestion control policy will be used. It indicates that receiver can accept more data but intermediate routers cannot.
  • If RW < CW, then sender window = RW and flow control policy will be used. So no need to set any threshold here.

Case 1 : CW = 100 MSS and RW = 16 MSS

so, sender window = 16 MSS. No congestion here. We can either use stop-n-wait or selective repeat flow control policy.

Using stop and wait (we send 1 MSS in 1 RTT) = 16*1 ms = 16 ms

Using selective repeat (entire window is sent during one RTT) = 1 ms

Case 2 : CW = 16 MSS and RW = 100 MSS

So, sender window = 16 MSS. Use congestion control policy. 

Since there is no timeout, no initial threshold is defined.

1st RTT => 1 MSS      ....slow start

2nd RTT => 2 MSS

3rd RTT => 4 MSS

4th RTT => 8 MSS

5th RTT => 16 MSS

So, we need 5 RTTs here to send full window which is 5 ms.

selected by
Answer:

Related questions

1 votes
1 votes
1 answer
2
Markzuck asked Feb 1, 2019
430 views
When during TCP congestion control, timeout occurs in 5th transition, then shall those packets be considered as received or shall be resent?Asking for if question asks fo...
2 votes
2 votes
1 answer
3