Transport Layer Services Overview
Uploaded by
Jithin STransport Layer Services Overview
Uploaded by
Jithin SCommon questions
Powered by AIThe transport layer is responsible for process-to-process delivery, ensuring data is sent between specific applications running on hosts using port numbers. It establishes end-to-end connections, using protocols like TCP for reliable communication with features such as sequence numbers, checksums, and acknowledgments to ensure data integrity. It also manages flow control (e.g., through TCP's sliding window protocol) and congestion control (e.g., TCP uses AIMD to adjust sending rates based on network conditions). Additionally, it performs multiplexing and demultiplexing to manage multiple data streams from different applications on a single host, and finally, it supports error correction by retransmitting lost or corrupt segments.
TCP (Transmission Control Protocol) is connection-oriented, requiring a three-way handshake to establish a reliable connection, involving sequence numbers, acknowledgments, and error correction through retransmissions. It handles congestion control using mechanisms like AIMD, ensuring reliable, ordered delivery, making it suitable for applications like web browsing and file transfers. UDP (User Datagram Protocol), on the other hand, is connectionless and lightweight, sending data without guaranteeing delivery or order. This results in minimal overhead (8-byte header), ideal for time-sensitive applications like DNS queries or live video streaming, where speed is prioritized over reliability. TCP’s robustness incurs more overhead, while UDP’s simplicity might result in data loss, requiring the application to manage reliability if needed.
UDP itself does not provide error handling or reliability features beyond a basic checksum for error detection, leaving these responsibilities to the application layer. Applications designed with UDP often implement custom mechanisms for handling lost packets and data integrity. For example, video streaming applications might use forward error correction (FEC) to recover lost data packets without retransmissions. Others might rely on redundancy or periodic status reviews to request only essential retransmissions. This approach allows applications to balance the need for speed and efficiency against reliability, thereby adapting to specific data requirements and network conditions.
The transport layer, particularly TCP, ensures data integrity through mechanisms like checksums, sequence numbers, and acknowledgments (ACKs). Each segment includes a 16-bit checksum calculated over its contents, checked at the receiver to determine if the data is corrupted. Sequence numbers allow tracking of segments, ensuring correct ordering and detection of any missing data. If a segment is lost or corrupted, TCP uses acknowledgments to confirm the receipt of data; missing ACKs signal the need for retransmission. In some protocols, negative acknowledgments (NACKs) are used to explicitly request missing data. These mechanisms together maintain data integrity by detecting errors and facilitating error correction through retransmission.
The transport layer uses port numbers to differentiate between multiple applications on the same host. Each application is assigned a unique port number within the 0–65,535 range, categorized into well-known ports (0–1023), registered ports (1024–49151), and ephemeral ports (49152–65535). For instance, a web server typically uses port 80, while a browser might use a dynamically assigned ephemeral port for communication. This combination of IP address and port number forms a socket that uniquely identifies each endpoint in a connection, allowing the transport layer to direct data to the correct application.
Using TCP for real-time applications can introduce latency due to its connection-oriented nature, requiring a handshake before data transfer and additional overhead for reliability features like acknowledgments and retransmissions. These features ensure complete and ordered data delivery, which is crucial for applications where data integrity is paramount, but they can disrupt the seamless flow needed in real-time applications like live video or voice calls. UDP, with its minimal latency owing to a connectionless setup, is often preferred for real-time applications despite lacking reliability, as such applications can tolerate some data loss in exchange for reduced latency and jitter, providing a more continuous stream of data.
Multiplexing, at the sender's side, combines data from different applications into a unified data stream, assigning each a unique port number and packaging them into segments tagged with source and destination ports. Demultiplexing occurs at the receiver's side, where segments are directed to the appropriate application by examining the destination port in the segment headers. The transport layer uses the socket concept (IP + port) to ensure efficient handling of multiple concurrent services, significantly improving network efficiency by allowing multiple applications to share a single network connection rather than requiring a dedicated connection for each. This capability is crucial for effective utilization of network resources and maintaining separation of data streams.
SCTP (Stream Control Transmission Protocol) combines features of TCP and UDP, providing reliable, connection-oriented data transmission like TCP, but also supporting multi-streaming and multi-homing for robustness and efficiency. It is used in telephony signaling and certain real-time applications. DCCP (Datagram Congestion Control Protocol) offers congestion control over datagram traffic similar to UDP but includes features to manage traffic congestion, suitable for real-time applications requiring rate adaptation. Unlike TCP, UDP remains simple, focusing on speedy delivery without congestion control or reliability, while protocols like SCTP and DCCP target specific needs like reliable multiplexing or congestion-controlled streaming.
TCP employs flow control using the sliding window protocol, where the receiver advertises its buffer capacity as a window size in ACKs. The sender respects this window size, limiting its data transmissions accordingly and pausing if the window size is zero, indicating that the receiver's buffer is full. As the receiver processes data and frees buffer space, the window slides forward, allowing the sender to continue. This dynamic adjustment prevents data loss due to buffer overflow and ensures that the sender does not transmit data faster than the receiver can process it.
The transport layer manages congestion control primarily through TCP, which employs strategies to prevent and react to congestion. Open-loop congestion control prevents congestion by shaping traffic, such as using the Leaky Bucket Algorithm to regulate data flow. Closed-loop congestion control allows TCP to react to network congestion through mechanisms like AIMD (Additive Increase, Multiplicative Decrease), which increases the sending rate gradually when no congestion is detected and decreases it multiplicatively (halves the sending rate) when congestion is apparent (detected through packet loss indicators like duplicate ACKs or timeouts). TCP also uses a dynamic congestion window (cwnd) to adjust the amount of data in transit based on current network conditions, ensuring efficient data transmission even in varied network environments.