Shared-hub bandwidth to a server: Ten users are connected to a classic hub operating at 10 Mbps half-duplex. The hub uplinks toward a switch, and a server on that switch also runs at 10 Mbps half-duplex. Assuming equal sharing on the collision domain, how much bandwidth does each host effectively have to the server?

Introduction / Context:
This question tests understanding of classic Ethernet hubs, collision domains, and half-duplex contention. Unlike a switch, a hub is a multiport repeater: all connected devices share the same 10 Mbps bandwidth and contend for access using CSMA/CD. Knowing how sharing works helps estimate per-host throughput to a server.

Given Data / Assumptions:

• Hub speed is 10 Mbps and half-duplex (collisions possible).
• Ten hosts are attached to the hub; traffic is destined to a 10 Mbps half-duplex server reachable through the hub’s uplink.
• Assume equal utilization and no QoS prioritization.

Concept / Approach:

On a hub, all ports form one collision domain and share the medium. Aggregate usable throughput among the contenders is bounded by 10 Mbps (less after CSMA/CD overhead). With roughly equal contention, a simple estimate divides the available rate by the number of active hosts.

Step-by-Step Solution:

Verification / Alternative check:

Packet captures show collisions and backoffs on a hub. Moving the users to a switch (separate collision domains per port) or increasing the server link rate changes the calculation. But with a hub, equal-share intuition matches lab measurements.

Why Other Options Are Wrong:

10 Mbps (D) would require a dedicated, non-contended link per host (a switch), which is not the case.

2 Mbps (C) and 100 kbps (A) do not match equal sharing of 10 Mbps among ten users.

Common Pitfalls:

Forgetting that hubs do not isolate collisions; assuming full-duplex or switched behavior; ignoring protocol overheads (which only reduce throughput further from the 1 Mbps estimate).

Final Answer:

1 Mbps