Network Protocols & Utilization Concepts

Delay and Utilization

• An 8000 bit packet takes 8 microseconds to be transmitted over a 10^9 bits/sec link.
• Sender utilization, the fraction of time a sender is busy sending is calculated as the packet size divided by the sum of round trip time and the time to transmit the packet.
• With a round trip time of 30 milliseconds and 1KB packets every 30 milliseconds, throughput is 33kB/sec over a 1Gbps link.
• Network protocols restrict the usage of physical resources.

Stop-and-Wait Operation

• The sender waits for an acknowledgment from the receiver before sending the next packet.
• Sender utilization in Stop-and-Wait is calculated as the packet size divided by the sum of round trip time and the time to transmit the packet, which is identical to the calculation for the initial delay example.

Pipelined Protocols

• Pipelined protocols allow multiple packets to be in transit at the same time.
• To support pipelining, the range of sequence numbers must be increased and buffering must be implemented at both the sender or the receiver.
• Pipelining increases the utilization of the network, for example 3-packet pipelining increases utilization by a factor of 3.

TCP Segment Structure

• TCP uses sequence numbers and acknowledgments to track the flow of data.
• Sequence numbers indicate the byte stream number of the first byte in a segment's data.
• Acknowledgments include the sequence number of the next byte expected from the other side and it serves as a cumulative acknowledgment.
• TCP specifies the window size, which indicates the amount of data the receiver is willing to accept.
• The receiver handles out-of-order segments by buffering them until the missing segments arrive.

TCP Round Trip Time and Timeout

• TCP uses round trip time (RTT) to track the time it takes for a packet to be transmitted and its acknowledgment to be received.

TCP Fast Retransmit

• TCP can retransmit a packet quickly if it receives three duplicate acknowledgments indicating that the packet has been lost.
• When a packet is retransmitted, a timeout is triggered.

Principles of Congestion Control

• Congestion occurs when too many sources send too much data too quickly for the network to handle.
• Congestion control aims to prevent congestion by dynamically adjusting the rate at which data is sent through the network.

TCP Fairness

• TCP aims for fairness by ensuring that each TCP connection sharing a bottleneck link receives approximately the same bandwidth.
• TCP uses additive increase multiplicative decrease (AIMD) to achieve fair bandwidth allocation among competing connections.
• Additive increase allows the throughput to increase linearly when there is no congestion.
• Multiplicative decrease reduces the throughput proportionally when congestion is detected.

TCP Flow Control

• TCP flow control regulates the rate at which a sender transmits data to the receiver.
• TCP uses a window mechanism to control the amount of data that can be sent at any time, ensuring the receiver's buffer does not become overwhelmed.

TCP 3-Way Handshake

• TCP establishes a connection using a 3-way handshake.
• The client initiates by sending a SYN (synchronization) packet to the server.
• The server responds with a SYN-ACK (synchronization-acknowledgment) packet.
• The client acknowledges the SYN-ACK with an ACK packet, completing the connection establishment.

TCP Closing a Connection

• Closing a connection occurs in a 4-way handshake.
• Both the client and server indicate their intention to close the connection by sending a FIN (finish) packet.
• The other party acknowledges the FIN with an ACK packet.
• The closing process concludes when both parties have received the FIN and ACK packets.

Principles of Congestion Control

• Congestion occurs when the network is overloaded with data.
• Congestion control reduces the amount of data in the network to avoid congestion collapse, and maximize throughput.

TCP Congestion Control

• TCP uses various mechanisms to control congestion through the network including slow start, congestion avoidance, fast retransmit, and fast recovery.

Fairness and UDP/TCP

• UDP applications do not use congestion control and therefore can suffer from high packet loss.
• Applications using UDP often send at a constant rate, tolerating packet loss for real-time streaming.
• Applications can open multiple TCP connections in parallel, leading to potential unfairness and bandwidth allocation issues.

Chapter 3: Summary

• The transport layer is responsible for providing services on top of the network layer.
• The transport layer handles multiplexing, demultiplexing, reliable data transfer, flow control, and congestion control.
• The transport layer is implemented in the Internet by two primary protocols: UDP (User Datagram Protocol) and TCP (Transmission Control Protocol).
• The transport layer is responsible for transmitting data through the network core, while the application layer handles the applications.