Note: Many topics at this site are reduced versions of the text in "The Encyclopedia of Networking and Telecommunications." Search results will not be as extensive as a search of the book's CD-ROM.

Network services can be categorized as best-effort, connectionless services or reliable connection- oriented services. In the Internet protocol suite, IP is a best-effort service and TCP is a reliable service. IP provides basic packet forwarding while TCP implements flow controls, acknowl- edgements, and retransmissions of lost or corrupted packets. This split in services "decentralizes" the network and moves the responsibility for reliable delivery to end systems. TCP is an end-to-end transport protocol, meaning that it runs in end systems, not the network. IP is a network protocol. This topic discusses flow control in terms of TCP, but flow controls are used by many other communication protocols, as mentioned later. The services offered by TCP include the following:

• Flow-control mechanisms control packet flow so that a sender does not transmit more packets than a receiver can process.


• Reliable delivery mechanisms provide a way for a receiving system to acknowledge that it has received a packet, and a way for the sender to know that it must retransmit a lost or corrupted packet. Refer to "Reliable Data Delivery Services."


• Congestion control mechanisms allow network systems to detect network congestion (a condition in which there is more traffic on the network than can be handled by the network or network devices) and throttle back their transmission to alleviate the congestion. Refer to "Congestion Control Mechanisms."

Congestion occurs on busy networks. When it occurs, end systems and the network must work together to minimize the congestion. In contrast, flow controls are used between end systems. A receiver uses flow controls to signal to the sender that it is overloaded. The sender then throttles back or stops its transmission.

Flow controls are necessary because senders and receivers are often unmatched in capacity and processing power. A receiver might not be able to process packets at the same speed as the sender. If buffers fill, packets are dropped. The goal of flow-control mechanisms is to prevent dropped packets that must be retransmitted.

Keep in mind that flow controls are used in the data link layer to control flow between devices that are directly connected. In contrast, TCP controls flow between devices that may be connected across a multihop routed network. Data link layer protocols include SDLC (Synchronous Data Link Control), HDLC (High-level Data Link Control), LAP-B (Link Access Procedure-Balanced), SLIP (Serial Line Internet Protocol), and PPP (Point-to-Point Protocol). Transport layer protocols include TCP (Transmission Control Protocol) and Novell SPX (Sequenced Packet Exchange).

When discussing data link layer protocols, the unit of transmission is the frame. In the transport layer, the unit of transmission for TCP is the segment. A segment is encapsulated in an IP datagram, which in turn is encapsulated in frames at the data link layer.

There are a few Internet RFCs related to flow control, including the original TCP RFCs:

• RFC 793 (Transmission Control Protocol, September 1981)


• RFC 813 (Window and Acknowledgement Strategy, July 1982)


• RFC 1016 (Something a Host Could Do with Source Quench: The Source Quench Introduce, July 1987)


• RFC 1180 (A TCP/IP Tutorial, January 1991)