Questions: Packet Switching vs. Circuit Switching

Question 1

100 users share a 1 Gbps packet-switched link. Each user can burst at 10 Mbps but is only actively transmitting 10% of the time. A network engineer claims this configuration works well in practice. What is the core reason?

• A. Each user is allocated a guaranteed 10 Mbps slice of the 1 Gbps link at all times
• B. Statistical multiplexing: because users are bursty and mostly idle, the probability that all 100 burst simultaneously is very low, so shared capacity suffices on average
• C. Packet switching compresses data, reducing the effective bandwidth required per user
• D. Routers distribute bandwidth perfectly equally so each user always gets exactly 10 Mbps

Answer: (B)

This is the fundamental efficiency argument for packet switching. If the link were circuit-switched, serving 100 users at 10 Mbps each would require 1 Gbps — exactly the link capacity — and idle time would be wasted. Packet switching uses statistical multiplexing: since real traffic is bursty and users are idle most of the time, the network can serve far more users than its dedicated capacity would suggest. When users don't all burst at once (which is statistically typical), everyone gets what they need. The cost of this efficiency is that when many users burst simultaneously, they experience queuing delays — variable latency is the price of statistical sharing.

Question 2

A surgeon is performing a remote operation over a network, relying on robotic arm feedback with a hard real-time requirement of sub-5ms round-trip latency at all times. Which switching approach better serves this use case?

• A. Packet switching — the internet's routing resilience ensures reliable delivery
• B. Circuit switching — it reserves a dedicated end-to-end path guaranteeing consistent, predictable latency without queuing variability
• C. Packet switching with QoS prioritization — always sufficient for real-time applications
• D. Circuit switching — because packets would arrive out of order and require reassembly delay

Answer: (B)

This is the scenario where circuit switching's key strength — guaranteed, predictable bandwidth and latency — matters most. Packet switching introduces variable latency because packets queue behind other traffic at routers; under load, delays can spike unpredictably. For a remote surgical system where a 50ms latency spike could cause physical harm, the guarantee of a dedicated circuit with no queuing variability justifies the inefficiency of reserved (sometimes idle) bandwidth. QoS can reduce latency variability in packet networks but cannot eliminate it entirely.

Question 3

In a packet-switched network, if a router on the primary path between two communicating hosts fails mid-session, the entire session is expected to be torn down and re-established from the beginning.

• T. True
• F. False

Answer: False

Fault tolerance is one of packet switching's core advantages. Because packets are independently routed and the network has no per-session state (unlike circuit switching's reserved paths), surviving routers can simply reroute subsequent packets along an alternate path. The receiving end reassembles packets using sequence numbers regardless of route. The session continues with a possible brief interruption but does not need to restart. This resilience was in fact a primary design goal of the original ARPANET — the predecessor to the internet.

Question 4

In circuit switching, reserved bandwidth for an active session remains unavailable to other connections even during periods when no data is being transmitted.

• T. True
• F. False

Answer: True

This is the defining inefficiency of circuit switching. When a circuit is established, resources (bandwidth on each link segment, switching capacity) are reserved end-to-end for the duration of the session. If you pause mid-phone-call, the circuit is still held open and no other call can use those reserved resources. Packet switching eliminates this waste through statistical multiplexing — shared links carry traffic from whoever has data to send at any moment, rather than sitting idle for a session that has nothing to transmit right now.

Question 5

Explain why the internet was designed around packet switching rather than circuit switching, given that packet switching introduces variable latency.

Model answer: Packet switching was chosen primarily for efficiency and resilience. Statistical multiplexing allows shared links to serve far more users than their raw capacity would permit under circuit switching, since real traffic is bursty rather than continuous. Packet switching also provides fault tolerance — if a router fails, packets reroute around it without session state being lost. The variable latency cost is acceptable for the vast majority of applications (web browsing, email, file transfer, streaming) that can tolerate some delay variability. For real-time applications requiring strict latency guarantees, modern networks use quality-of-service mechanisms to reduce (though not eliminate) variability.

The key insight is that variable latency is a tradeoff, not a flaw. For most use cases, the efficiency and resilience gains of packet switching far outweigh the cost of managing variable delay. The internet's success proves this empirically — virtually all global data communication now runs on packet-switched infrastructure, with circuit switching surviving only in specialized contexts like legacy telephony and some industrial control systems.