Question

Suppose two TCP connections share a path through a router R. The router's queue size is six segments; each connection has a stable congestion window of three segments. No congestion control is used by these connections. A third TCP connection now is attempted, also through R. The third connection does not use congestion control either. Describe a scenario in which, for at least a while, the third connection gets none of the available bandwidth, and the first two connections proceed with \(50 \%\) each. Does it matter if the third connection uses slow start? How does full congestion avoidance on the part of the first two connections help solve this?

Short answer

The third connection gets no bandwidth initially because the queue is filled by the first two connections. Slow start for the third connection doesn't help without room in the queue. Full congestion avoidance by the first two connections allows fairer bandwidth sharing.

Step-by-step solution

Understand the queue and congestion windows

The router R has a queue size of six segments. Each of the two initial connections has a congestion window of three segments. This means that under full load, both connections together would fill the router's queue exactly (3 segments per connection * 2 connections = 6 segments).

Introduce the third connection

A third TCP connection attempts to use the router R. This new connection also does not use congestion control. Since the router queue is already full with the segments from the first two connections, there is initially no space for the third connection's segments.

Identify bandwidth allocation without congestion control

Without congestion control, the first two connections continue to fill the router's queue with their segments. As segments are processed and leave the queue, the next segments from the first two connections will immediately fill the available space. Therefore, the third connection will get none of the available bandwidth because its segments cannot enter the router's queue.

Analyze the impact of slow start for the third connection

If the third connection uses slow start, it starts with a smaller congestion window and gradually increases it. However, since the first two connections already fill the router queue without ever leaving room for the third connection (because they quickly refill any emptied space), the third connection's slow start does not help it gain access to the queue or bandwidth.

Examine the effect of full congestion avoidance

If the first two connections use full congestion avoidance, they respond to congestion signals by reducing their window sizes. This would allow space in the router's queue periodically, giving the third connection a chance to place its segments in the queue and share bandwidth more fairly.

Key concepts

Router Queue Management

The router in the exercise has a queue size of six segments. This queue management is crucial for handling the packets that flow through it. When the router's queue is full, new packets cannot enter until space is available. This scenario leads to the dropping of packets or delayed transmission, often resulting in congestion.
Effective router queue management aims to balance the flow of packets. This balance helps to avoid congestion and ensure fair bandwidth allocation. Routers may use algorithms to dynamically manage the queue, allowing for better handling of fluctuating network demands.

Congestion Window

The congestion window is an essential part of TCP's congestion control mechanism. It dictates the number of segments a TCP sender can transmit without receiving an acknowledgment from the receiver. In the provided exercise, each of the initial two TCP connections has a congestion window of three segments.

When the congestion window is high, more data can be sent before awaiting acknowledgment. However, if congestion occurs (when the network can't handle the volume), the congestion window size will reduce to avoid overwhelming the network, resulting in fewer segments being sent.

Ultimately, the congestion window helps balance the data flow based on current network conditions, ensuring the network does not become overloaded.

Bandwidth Allocation

Bandwidth allocation refers to how network bandwidth is distributed across various connections. In the exercise scenario where no congestion control is applied, the first two connections dominate the available bandwidth by quickly filling up the router's queue.

The third connection fails to get any bandwidth because the available space in the queue is immediately filled by the first two connections' segments. This unfair distribution highlights the importance of incorporating proper congestion control techniques.

• With congestion control, each connection responds to network congestion, leading to more equitable bandwidth sharing.
• Without it, some connections may monopolize the bandwidth, leaving others without any.

Slow Start Mechanism

Slow start is a phase in TCP congestion control designed to prevent congestion before it happens. When a connection starts, it begins with a small congestion window and gradually increases it as acknowledgments are received.

During slow start, the congestion window grows exponentially until it reaches a threshold, after which it grows linearly. This gradual increase helps to probe the network's capacity and find an optimal data transmission rate.

In the scenario, even if the third connection uses slow start, it would still struggle to transmit data as the initial window size is too small to compete with the already full router queue from the first two connections.

Congestion Avoidance

Congestion avoidance is another phase in TCP congestion control aimed at maintaining efficient network usage without causing congestion. It typically follows the slow start phase once a certain threshold is reached.

In congestion avoidance, the congestion window size increases linearly, allowing the connection to adjust its rate of data transmission to the network conditions more finely. If the network starts showing signs of congestion, the window size is reduced to prevent further congestion.

In the exercise scenario, if the first two connections used full congestion avoidance, they would not fill the router queue continuously. This would leave periodic gaps in the queue, allowing the third connection a chance to transmit its segments and ensuring a fairer bandwidth allocation across all connections.