You are on page 1of 3

Chapter 5: Transport Layer The Need for a Transport Layer o Additional processing is required before the data-transfer (network

k Layer) o Three vital tasks (bookkeeping functions) Segmentation- chopping larger blocks of data into segments before going to the network layer Reliability- resolves the lost and duplicated packets Multiplexing- allows multiple applications on a computer to share a common network link- distributing incoming data packets to the appropriate application o The sending end of the transport layer receives data from application layer and brakes it into segments, it keeps track of the segments to account for duplicate/missing packets o Provides mechanism to distinguish between data segments on each application o Provides for graceful use of network resources by allowing receiving computers to specify the data transfer rate o Transmission Control Protocol (TPC)- most popular protocol- highly reliable host-to-host transport-layer protocol over packet-switched networks (end-users email and web) o User Datagram Protocol (UDP)- simpler transport-layer for applications that do not need segmentation and reliability- eliminates computational processing associated with TPC Transmission Control Protocol o Hands data that needs to be transmitted over a network (sending end) Sending end- breaks data into smaller segments (65 KB data limit for IP) then assigns sequence numbers to each segment and hands over the segments and IP transmits them to receiving computers Receiving end- sequence numbers are used to reassemble data to original form then passes reconstructed data to the application and user application unaware o Provides reliable communication service between applications over networks that are complex and levels of unreliability TCP Functions o Segmentation: Allows IP to transfer application data of arbitrary size- assists by breaking application data into segments of manageable size called datagrams Data packet names: (specifies layers) Frames- data created by data-link layer Packets- IP creates packets in the network layer Datagrams- TCP and UDP create diagrams Adds sequence number to every byte diagram before transmission- helps to reorder datagrams even if received out of order Assigned sequence numbers are calculated as the sequence of previous segment + length of previous segment (first is 1, second is 1+previous length) Advantages: scheme makes acknowledgements easier- TCP generally acknowledges datagrams by datagram boundary, and partial datagrams

Advantages: Reduces chance for error Reduces the need for reprocessing in case of error Easier for routers to handle segments Simplifies packet handling Knock-down kits- furniture reassembly- easy to handle, take less space, lowers cost Disadvantages: Additional complexity introduced by segmentation and reassembly- loss of segments or duplication Reliability: Mechanism for getting reliability is for the receiver to periodically acknowledge received datagrams- is not received in reasonable time, assumes worst resends Time-out interval- waiting time for sender to receive the segment Second dimension of reliability- ensures received datagrams are complete Fragmentation function of IP allows further fragmentation of packets to fit the Ethernet frame (1,500 bytes) IP only detects header errors, not packet-data errors. Computes a checksum to each diagram Multiplexing Combination of two or more information channels onto a common transmission medium Allows multiple applications to share the same network interface card and network link The information flowing to and from each individual application is an informational channel and the common transmission medium is the network connection of the computer TCP enables multiplexing by providing multiple port addresses within each host Network IP address and a port address together constitute a communication socket (page 129) How does the client PC know the IP address of the UCA file server? How does the client know that the UCA DNS server is running on which port? DNS is a service that translates site names such as www.X.edu to IP addresses such as 163.31.3.35 How do we know what port to connect to? Client port number is assigned by the local operating system IANA- Internet Assigned Numbers Authority defines standard portswhen a port is not specified, it connects to standard port o Standard port for web servers is 80 o Standard port for email is 25 o Standard port for windows file sharing is 445

An error message will occur if two different applications on the same port and directed to the web server and not your application o Avoid reserved ports (0-1023) for your own applications

You might also like