Systems Approach to Computer Networks Unit 8 ReviewTransport Layer: Reliable Data Transfer

Key Concepts

• Reliable data transfer ensures data is delivered accurately and in the correct order from sender to receiver
• Focuses on handling issues such as packet loss, corruption, duplication, and reordering that can occur during network transmission
• Utilizes techniques like error detection, error correction, sequence numbering, and acknowledgments to achieve reliability
• Implements flow control mechanisms to prevent the sender from overwhelming the receiver with data
  • Manages the rate at which data is sent based on the receiver's capacity to process and store the incoming data
• Incorporates congestion control strategies to avoid network congestion and maintain optimal performance
• Builds upon the services provided by the underlying network layer (typically IP) to provide a reliable end-to-end communication channel
• Commonly implemented in the transport layer protocols such as TCP (Transmission Control Protocol)

Reliable Data Transfer Protocols

• Implement specific algorithms and mechanisms to ensure reliable data delivery between sender and receiver
• Use sequence numbers to identify and order packets, allowing the receiver to detect missing or out-of-order packets
• Employ acknowledgments (ACKs) sent by the receiver to confirm the successful receipt of packets
  • Positive acknowledgments (ACKs) indicate correctly received packets
  • Negative acknowledgments (NACKs) indicate corrupted or missing packets
• Utilize timeouts and retransmissions to handle lost or delayed packets
  • If an acknowledgment is not received within a specified time (timeout), the sender assumes the packet was lost and retransmits it
• Implement error detection techniques (checksums, CRCs) to identify corrupted packets and request retransmission
• Incorporate sliding window protocols to enable efficient and reliable data transfer
  • Allow multiple packets to be sent and acknowledged simultaneously, improving throughput

Challenges in Network Communication

• Packet loss occurs when packets fail to reach their destination due to network congestion, link failures, or other factors
  • Results in missing data and requires retransmission of lost packets
• Packet corruption happens when bits in a packet are altered during transmission, leading to invalid data
  • Can be caused by physical layer issues (noise, interference) or hardware malfunctions
• Packet duplication arises when the same packet is received multiple times by the receiver
  • May occur due to retransmissions or network anomalies, requiring duplicate detection and handling
• Packet reordering happens when packets arrive at the receiver in a different order than they were sent
  • Can be caused by multiple paths, varying network delays, or parallel processing at intermediate nodes
• Limited bandwidth and network congestion can impact the performance and reliability of data transfer
• Heterogeneous network environments with different link capacities, latencies, and reliability characteristics pose challenges

Error Detection and Correction

• Error detection techniques identify errors in transmitted data without necessarily correcting them
  • Checksums calculate a fixed-size value based on the data and append it to the packet for verification at the receiver
  • Cyclic Redundancy Checks (CRCs) use polynomial division to generate a checksum that can detect a wide range of errors
• Error correction techniques not only detect errors but also attempt to correct them without retransmission
  • Forward Error Correction (FEC) adds redundant data to the original message, allowing the receiver to recover from a limited number of errors
  • Automatic Repeat Request (ARQ) combines error detection with retransmission of erroneous packets
    • Stop-and-Wait ARQ sends one packet at a time and waits for an acknowledgment before sending the next packet
    • Go-Back-N ARQ uses a sliding window to send multiple packets and retransmits all packets starting from the lost or corrupted one
    • Selective Repeat ARQ retransmits only the specific packets that are lost or corrupted, minimizing unnecessary retransmissions

Flow Control Mechanisms

• Prevent the sender from overwhelming the receiver by sending data faster than the receiver can process and store it
• Sliding window protocol is a common flow control mechanism used in reliable data transfer protocols
  • Sender maintains a window of packets that can be sent without waiting for individual acknowledgments
  • Receiver advertises its available buffer space (receiver window) to the sender, limiting the amount of unacknowledged data
• Stop-and-Wait flow control sends one packet at a time and waits for an acknowledgment before sending the next packet
  • Simple but inefficient for high-latency networks due to the idle time between packets
• Credit-based flow control assigns credits to the sender, representing the number of packets the receiver can accept
  • Sender decrements credits as packets are sent and increments them as acknowledgments are received
• Rate-based flow control adjusts the sending rate based on the receiver's processing capacity and network conditions
  • Receiver provides feedback to the sender about its current receiving rate or available buffer space

Congestion Control Strategies

• Aim to prevent and mitigate network congestion by regulating the rate at which senders inject data into the network
• Slow start is a congestion control mechanism used by TCP to gradually increase the sending rate at the beginning of a connection
  • Sender starts with a small congestion window (cwnd) and doubles it with each round-trip time (RTT) until a threshold is reached
• Congestion avoidance phase follows the slow start and aims to find the optimal sending rate without causing congestion
  • Sender increases the congestion window by one maximum segment size (MSS) per RTT until congestion is detected
• Fast retransmit and fast recovery optimize the recovery process when packet loss is detected through duplicate acknowledgments
  • Sender retransmits the lost packet immediately and reduces the congestion window by a factor (usually half) to alleviate congestion
• Explicit Congestion Notification (ECN) allows routers to explicitly signal congestion to the sender by marking packets
  • Sender responds to ECN signals by reducing its sending rate to prevent further congestion
• TCP congestion control algorithms (Reno, NewReno, Cubic) implement different strategies to adapt the sending rate based on network conditions

TCP vs UDP: A Comparison

• TCP (Transmission Control Protocol) provides reliable, ordered, and error-checked delivery of data between applications
  • Establishes a connection-oriented communication channel before data transfer begins
  • Guarantees that data is delivered accurately and in the same order as sent
  • Implements flow control, congestion control, and error recovery mechanisms
  • Suitable for applications that require reliable data delivery (web browsing, email, file transfer)
• UDP (User Datagram Protocol) offers a connectionless, unreliable, and best-effort delivery service
  • Does not establish a prior connection and sends data without any reliability guarantees
  • Provides minimal error checking and does not ensure ordered delivery or retransmission of lost packets
  • Has lower overhead and latency compared to TCP due to its simplicity
  • Suitable for applications that prioritize speed and can tolerate some data loss (streaming media, online gaming, DNS)

Real-World Applications

• Web browsers use TCP to ensure reliable delivery of web pages, images, and other content from servers to clients
• Email clients and servers rely on TCP to guarantee the accurate and complete transfer of email messages and attachments
• File transfer protocols (FTP, SFTP) employ TCP to ensure the integrity and completeness of transferred files
• Voice over IP (VoIP) applications often use UDP for real-time audio transmission, prioritizing low latency over reliability
• Online gaming platforms commonly utilize UDP for fast and responsive gameplay, tolerating occasional packet loss
• Streaming media services (video streaming, online radio) may use UDP or adaptive bitrate streaming over TCP to optimize performance
• Domain Name System (DNS) primarily uses UDP for fast and efficient resolution of domain names to IP addresses
• Remote administration tools (SSH, RDP) rely on TCP for secure and reliable communication between remote systems