DOC PREVIEW
Yale CPSC 155 - TCP And UDP

This preview shows page 1 out of 4 pages.

Save
View full document
View full document
Premium Document
Do you want full access? Go Premium and unlock all 4 pages.
Access to all documents
Download any document
Ad free experience
Premium Document
Do you want full access? Go Premium and unlock all 4 pages.
Access to all documents
Download any document
Ad free experience

Unformatted text preview:

TCP And UDP By Steve Steinke, Network Magazine Feb 5, 2001 (10:03 AM) URL: http://www.networkmagazine.com/article/NMG20010126S0005 The two Transport layer protocols in the TCP/IP family, TCP and UDP, provide network services for applications and Application layer protocols (including HTTP, SMTP, SNMP, FTP, and Telnet.) These two protocols perform those services by employing the IP to route packets to their destination networks. TCP provides connection-oriented, reliable, byte-stream packet delivery, while UDP provides connectionless, unreliable, byte-stream packet delivery. These terms need explanation. Connection-oriented protocols establish an end-to-end link before any data moves. ATM and frame relay are connection-oriented protocols, but they operate at the Data-link layer rather than the Transport layer. Placing an ordinary voice phone call is also connection-oriented. Reliable protocols safeguard against several forms of transmission mishaps. You can compare checksums included with a packet's data payload with a recalculation of the checksum algorithm at the destination to detect corrupted data. You must retransmit corrupted or lost data, so the protocol must provide methods for the destination to signal the source when retransmission is needed. Packetized data can arrive out of sequence, so the protocol must have a way to detect out-of-sequence packets, buffer them, and pass them to the Application layer in the correct order. It must also detect and discard duplicate transmissions. A collection of timers enables limiting the wait for various acknowledgements, so you can initiate retransmissions or link re-establishment. Byte-stream protocols don't specifically support data units other than bytes. TCP can't structure bytes of the data payload in a packet, nor can it cope with individual bits. As far as TCP is concerned, it's responsible for transporting an unstructured string of 8-bit bytes. A connectionless protocol doesn't establish paths across the network before data can flow. Instead, the protocol routes connectionless packets or datagrams individually at each intermediate node. Without an end-to-end link, a connectionless protocol such as UDP isn't reliable. When a UDP packet moves into the network, the sending process can't know whether the packet arrives at its destination unless the Application layer acknowledges this fact. Nor can the protocol detect duplicate or out-of-sequence packets. The standard jargon describes UDP as “unreliable,” though a more descriptive term might be “nonreliable.” On modern networks, UDP traffic isn't prone to disruption, but you can't really call it “reliable,” either.Figure 1 shows the fields of a TCP segment, the part of an IP packet that follows the IP header information. The first 16 bits identify the source port, and the second 16 bits identify the destination port. Port numbers provide a way for IP hosts to multiplex numerous types of concurrent connections at a single IP address. The combination of a 32-bit IP address and a 16-bit port address identifies a socket in most modern operating systems. The combination of a source socket and a destination socket defines a TCP connection. There are 216 or 65,536 possible ports. The lowest 1,024 ports are called well-known ports; these are set aside by default for particular Application layer protocols. For example, HTTP uses port 80 by default, while POP3 uses port 110. Other applications can use the higher port numbers. The next two fields, the sequence number and the acknowledgement number, are the keys to TCP's reliability functions. When a TCP connection is established, the initiating host sends an arbitrary initial sequence number to the initiatee. The initiatee adds 1 to the sequence number and returns it to the initiator in the acknowledgement field, thereby indicating the next byte that should be sent. Once data begins to flow, the sequence and acknowledgement numbers keep track of which data bytes have been sent and which data bytes have been acknowledged. Because each field is 32 bits, it can have 232 values, so each field ranges from 0 to 4,294,967,295 and wraps around to 0 when it passes the upper limit. The 4-bit data offset field simply indicates how many 32-bit words the TCP header has. This information is necessary because there are optional header fields, and the data offset marks where the header ends and the data begins. TCP designers set aside the next 6 bits, just in case they might be needed for future development. Since RFC793 (Transmission Control Protocol) dates to 1981, and no one since has established a good reason to use these reserved bits, Jon Postel and his colleagues must have been overly cautious. Each of the following 6 bits is a flag. An URG (urgent) flag with a value of 1 indicates that the data in the urgent pointer field, farther along in the header, is significant. An ACK (acknowledgement) flag with a value of 1 indicates that the data in the acknowledgement number field is significant. (Note that an initial setup or SYN packet has a meaningful sequence number, but not a meaningful acknowledgement number, since it isn't acknowledging anything.) The PSH (push) flag prevents data from waiting to be sent and from waiting to be processed by the receiving process. The RST (reset) flag shuts down a connection. The SYN (synchronization) flag indicates that the sequence number is significant. The FIN (finish) flag indicates the sender has no more data to send. The Window field, 16 bits long, indicates the size of a so-called sliding window, which tells the sender how many bytes of data it's prepared to accept. TCP controls flow and congestion by adjusting the window size. A window equal to 0 tells a sender the receiver is overwhelmed and can't accept anything more without further notice. Large window sizes enable as many as 65,536 unacknowledged bytes to be in transit at any given time, but congestion—indicated when the retransmission timer expires without an acknowledgement—cuts the window size in half, effectively slowing the transmission rate. The 16-bit checksum field protects the data payload's integrity, the TCP header, and certain fields of the IP header. The sender calculates the checksum value and inserts it in this field, and the receiver recalculates the value based on the received packet and compares the two. If they match, the data is probably intact.The urgent pointer is a 16-bit offset value indicating the last


View Full Document

Yale CPSC 155 - TCP And UDP

Documents in this Course
Portals

Portals

16 pages

Lecture 4

Lecture 4

30 pages

Lecture 2

Lecture 2

20 pages

Lecture 2

Lecture 2

19 pages

Lecture 9

Lecture 9

25 pages

Lecture 7

Lecture 7

23 pages

Load more
Download TCP And UDP
Our administrator received your request to download this document. We will send you the file to your email shortly.
Loading Unlocking...
Login

Join to view TCP And UDP and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view TCP And UDP 2 2 and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?