TCP/IP Protocol Suite
Brad Kauble
Benjamin Miller
Networking 101 Section 104
November 20, 2013

In the world of the Internet and data networking, applications need a way to transfer information form one user to another, anywhere, anytime in a manner that does not disrupt the applications usefulness. Devices connected to a network use multiple applications simultaneously that send and receive information. That data cannot just be sent across the network with any hope of getting to its destination in any recognizable state if it were not for the Transport layer. The Transport layer divides the data into manageable pieces, packaging and addressing them so they will arrive at their proper destination and then reassembling the pieces into the original signal or stream. This paper will observe the part the Transport layer plays in application data encapsulating for use by the Network layer; how it allows simultaneous communication of multiple applications on a single host device while ensuring reliable data delivery to the right application correcting errors if required.
The Transport layer is layer 4 in the Open Systems Interconnection (OSI) Model and layer 2 in the Transmission Control Protocol/Internet Protocol (TCP/IP) Model. The Transport layer is needed to track individual conversations, segment, manage and reassemble data packets, identify sending and receiving applications, perform flow control between end users, enable error recovery and initiate and terminate sessions. Any one host may have several applications that are using the network to communicate at one time. The Transport layer’s responsibility is to keep track of and maintain the communication streams that are taking place with each of these applications.
Each application creates a stream of data that it uses to communicate with another application using the network. Most of these streams cannot be sent across the network all at once so the transport layer uses protocols to segment and encapsulate those data streams. Each data segment needs a way to identify where the data originated, where it is going, where it belongs in the data stream and to what application it is associated. The transport layer adds a header to accomplish this task. At the receiving end of the communication the data segments are reconstructed into a useful data stream using the information stored in the header, which identifies the target application. To do this, each application is assigned an identifier called a port number that is unique to each host. The Transport layer uses the port number to identify the application where each data segment is associated.
Applications use different requirements for their data. Some applications require that all of the data to be received before it can be used while others can tolerate missing bits here and there during transmission, like streaming video and voice over IP (VoIP). Still others require that the data segments be received in order, like email and web browsing. The Transport layer provides multiple protocols to handle all of these varying application data requirements. Each protocol has different rules to handle the different requirements of each application. These protocols separate all of the different simultaneous communications that my take place on a host computer connected to a network such as sending and receiving email, web browsing, VoIP, streaming music or video and instant messaging. The Transport layer protocols insure that each communication segment reaches the desired destination application in the manner that it is required.
The Transport Layer controls the conversations by creating connection sessions between applications. This connection creation prepares applications for communication before any data is transmitted. It also insures that if a piece of needed data is lost or corrupted during transmission it can be resent. Since there are multiple pathways that data can physically travel, some of it can arrive at the destination out of order. Data sequencing ensures the Transport layer can reassemble the data in the correct order. Network hosts may have limited bandwidth or memory and cannot accept the amount of data being sent at one time. Some protocols can change the rate of data that is sent. This reduces the need for data retransmission. Data reliability versus speed of data delivery becomes a concern for application developers. They need to decide which transport protocol will work best for their application across the network.
Web pages, email and database applications require that all of the data in the stream be received in the correct order that it was sent to be useful. These applications need reliable communication for the applications to work properly. This reliability requirement increases overhead over the network and slows communication. Conversely, streaming video and VoIP are not as sensitive to missing data packets in the data stream.
This leads to the two most common TCP/IP protocols, Transmission Control Protocol (TCP) and User Data Protocol (UDP). Both of these protocols can be used by multiple applications and each has specific functions that they implement.
TCP offers a controlled and connected conversation. The control adds overhead to gain functionality but offers reliable and same order delivery with flow control. TCP adds a 20-byte header encapsulating the Application layer data segment. File transfers, email and browsers all use TCP.
UDP is simple, uncontrolled and unconnected but offers speed as a tradeoff. It has a low overhead 8-byte header. Video streaming, VoIP and Domain Name System (DNS) are applications that use UDP.
The Transport layer identifies conversations by the source and destination port number. The header of the datagram or segment contains both the source and destination port numbers. Port number assignment varies. Servers have static port numbers that do not change, client’s request port numbers are chosen at random for each conversation. Many common applications have default or “well-known” port numbers. Well-known port numbers have been assigned the lowest group ranging from 0 to 1023; registered port numbers have been assigned the range from 1024 to 49151; and dynamic or private ports are assigned the range from 49152 to 65535 by the Internet Assigned Numbers Authority (IANA). The following table illustrates some Well Known and Registered ports, their applications and the protocol that the application uses.

Well-known Port Application Protocol

20 File Transfer Protocol (FTP) TCP 21 File Transfer Protocol (FTP) TCP 23 Telnet TCP 25 Simple Mail Transfer Protocol (SMTP) TCP 53 Domain Name System (DNS) TCP/UDP 69 Trivial File Transfer Protocol (TFTP) UDP 80 Hypertext Transfer Protocol (HTTP) TCP 110 Post Office Protocol (POP) TCP 161 Simple Network Management Protocol (SNMP) TCP/UDP 194 Internet Relay Chat (IRC) TCP 443 Secure HTTP (HTTPS) TCP 520 Routing Information Protocol (RIP) UDP

Registered Port Application Protocol
1433 MS Structured Query Language (MS SQL) TCP/UDP
1812 RADIUS Authentication Protocol UDP
1863 MSN Messenger TCP
2000 Cisco Skinny Client Control Protocol UDP (SCCP, used in VoIP applications)
2948 Wireless Application Protocol (WAP) TCP/UDP
5004 Real-Time Transport Protocol (RTP, a UDP voice and video transport protocol)
5060 Session Initiation Protocol (SIP, used in UDP VoIP applications)
8008 Alternate HTTP TCP
8080 Alternate HTTP TCP
In order for two hosts to communicate with each other using TCP they need to establish a connection. This is done using a three-way handshake. The initiating server host sends a data segment containing a random Initial Sequence Number (ISN) value and the SYN (Synchronize Sequence Number) control flag set. Once the SYN is received the responding client host increments the initial value and sends back a reply with the acknowledgement control flags SYN and ACK set. When the server receives the acknowledgement it increments the initial value and replies with only the ACK control flag set completing the process of establishing communication. At this time data can now be transferred between hosts across the network using the preferred application protocol.
If any data is lost during transmission the receiver cannot acknowledge its receipt. When the sender does not receive the acknowledgement for the data sent, then the sender can resend the missing data segments so they can be reassembled in the correct order at the receivers end and complete the conversation.
When data transfer is complete, the session must be terminated. To do this, TCP uses a two-way handshake to end each one-way TCP communication. The client, or data sender, sends a segment with the FIN (finish) control flag set. The server sends an ACK to acknowledge the receipt of the FIN that terminates the communication between the client and the server. The server then sends a FIN to the client who replies with and ACK to close the communication between the server and the client. Communicating using UDP, hosts do not need to establish a connection before sending data. When an application has data to send, it sends the data. The data is segmented and sent across the network and reassembled on the other end as it is received. Datagrams may not be in order and some may be missing entirely but are not resent. Like TCP, UDP applications servers are assigned Well Known or Registered port numbers so as UDP receives datagrams they are sent to their assigned port number. When a client application requests data from a server process the client is assigned a random source port number and the destination port will be Well Known or Registered. The random source port assignment helps keep the client more secure because intruders won’t know what client ports are open. In conclusion the Transport layer is needed to divide data from applications into smaller segments to be sent across the network, managing each data segment by adding a header, reassembling the data segment and delivering the segment to the correct application using the information in the header. The common protocols are TCP and UDP.
TCP data segments and UDP datagrams have headers that contain their source and destination port numbers that allow the data to be set to the correct application. Applications that use TCP and UDP are either Well Known or Registered. Requesting clients are assigned a random source port number, which adds to their security.
TCP creates a connection prior to passing any data using a three-way handshake. It manages the flow of data resending data segments that are lost or corrupt and controls the flow of data by adjusting the data rate. TCP is a reliable method of transferring data but it has larger segment headers, which slows down the transmission speed.
UDP on the other hand does not need to create a connection and it does not keep track of the data being transmitted. It sends and delivers data as it is received to the correct port destination. UDP uses a smaller datagram header and is less reliable than TCP but it has a significant higher data transfer rate.

Bibliography
Krzysztof Iniewski, C. M. (2007). TCP/IP Protocol Suite. Wiley Online Library.
Mateti, P. (2007, August 15). Transmission Control Protocol (TCP)/Internet Protocol (IP) Suite.
Wrigt, G. R., & Stevens, W. R. (2004). TCP/IP Illustrated Volume 2. Boston: Addison-Wesley Professional.