0

Studying Packet Switching. Encountered a question - enter image description here

What i know?

  1. Packets are transmitted over each communication link at the rate equal to full transmission rate of the link.
  2. Router is using Store-and-Forward Transmission.
  3. Overall throughput will be the throughput of the bottleneck link i.e R3

Problem?

If router is using Store-and-Forward Transmission then the Link R1 should be utilised in full i.e. 100 Mbps (1.0 in answer to the question) and overall throughput should be 0.10 but answer to Link Utilisation is given 0.10. Am I missing something?

Background?

This question is from the basic introductory chapter of the book I am referring, hence please ignore any kind of delays and other relevant algorithms like processing, queuing, ack etc.

1 Answer 1

0

If router is using Store-and-Forward Transmission then the Link R1 should be utilised in full i.e. 100 Mbps (1.0 in answer to the question) and overall throughput should be 0.10 but answer to Link Utilisation is given 0.10. Am I missing something?

You are ignoring that the router's buffer is small compared to the (assumed) total volume of traffic. If 1 MB is buffered (which is rather high) then that buffer is exhausted in roughly .1 seconds (1 MB / (8 bit/B) / (100 Mbit/s - 10 Mbit/s)).

Most of the time, a controlled flow as with TCP is assumed that doesn't start at full speed but ramps up during the initial phase, so buffer saturation is somewhat slower and takes a bit longer. Still, the buffer fills up and loses its capability to allow asymmetric flow.

Of course, with an uncontrolled flow like with UDP the left source host could send at its full link speed and the router would drop 90% of that traffic (except for initial buffering).

please ignore any kind of delays and other relevant algorithms like processing, queuing, ack etc

Ignoring ACK you'd get an uncontrolled flow. Since most of the traffic is lost in that scenario it's not really useful to do so.

2
  • Thank you for the answer. If I wish to summarise - My answer 1.0 will be correct if somehow (hypothetically) we have infinite buffer and every single packet is delivered without any error. The answer 0.1 is correct practically considering the limited buffer and controlled flow. Right?
    – Mod
    Aug 27, 2022 at 8:58
  • 1
    The buffer doesn't need to be infinite, but it'd need to be able to store the entire flow. But you're correct, yes. Since buffering actually works much better with smaller buffers (see "buffer bloat"), flows have unknown length, and TCP's congestion control works to avoid buffer bloat, it is commonly assumed that buffers are too small to actually cause the "asynchronous" flow you describe. And that book makes a lot of assumptions.
    – Zac67
    Aug 27, 2022 at 9:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.