tcp/ip overview 14th nov' 2007

course: refresher course on intranet applications with java
venue: academic staff college, jntu kukkatpally, hyderabad


TCP/IP
Transmission Control Protocol / Internet Protocol
• Rajesh Kulkarni
TCP/IP & OSI
• In OSI reference model terminology -the TCP/IP protocol suite covers the network and transport layers.

• TCP/IP can be used on many data-link layers (can support many network hardware implementations).
Internet Protocol
The IP in TCP/IP
• IP is the network layer
• packet delivery service (host-to-host).
• translation between different data-link protocols.
IP Datagrams
• IP provides connectionless, unreliable delivery of IP datagrams.
• Connectionless: each datagram is independent of all others.
• Unreliable: there is no guarantee that datagrams are delivered correctly or even delivered at all.
IP Addresses
• IP addresses are not the same as the underlying data-link (MAC) addresses.

Why ?
IP Addresses
• IP is a network layer - it must be capable of providing communication between hosts on different kinds of networks (different data-link implementations).
• The address must include information about what network the receiving host is on. This is what makes routing feasible.
IP Addresses
• IP addresses are logical addresses (not physical)
• 32 bits.
• Includes a network ID and a host ID.
• Every host must have a unique IP address.
• IP addresses are assigned by a central authority (American Registry for Internet Numbers for North America).

The four formats of IP Addresses

Class A
 128 possible network IDs
 over 4 million host IDs per network ID
Network and Host IDs
• A Network ID is assigned to an organization by a global authority.

• Host IDs are assigned locally by a system administrator.

• Both the Network ID and the Host ID are used for routing.
IP Addresses
• IP Addresses are usually shown in dotted decimal notation:
1.2.3.4 00000001 00000010 00000011 00000100
• cs.rpi.edu is 128.213.1.1
10000000 11010101 00000001 00000001
Host and Network Addresses
• A single network interface is assigned a single IP address called the host address.
• A host may have multiple interfaces, and therefore multiple host addresses.
• Hosts that share a network all have the same IP network address (the network ID).
Subnet Addresses
• An organization can subdivide it’s host address space into groups called subnets.
• The subnet ID is generally used to group hosts based on the physical network topology.
Subnetting
Subnetting
• Subnets can simplify routing.
• IP subnet broadcasts have a hostID of all 1s.
• It is possible to have a single wire network with multiple subnets.
Mapping IP Addresses to Hardware Addresses
• IP Addresses are not recognized by hardware.
• If we know the IP address of a host, how do we find out the hardware address ?
• The process of finding the hardware address of a host given the IP address is called
Address Resolution
Reverse Address Resolution
• The process of finding out the IP address of a host given a hardware address is called
Reverse Address Resolution

• Reverse address resolution is needed by diskless workstations when booting (which used to be quite common).
ARP
• The Address Resolution Protocol is used by a sending host when it knows the IP address of the destination but needs the Ethernet (or whatever) address.
• ARP is a broadcast protocol - every host on the network receives the request.
• Each host checks the request against it’s IP address - the right one responds.
ARP (cont.)
• ARP does not need to be done every time an IP datagram is sent - hosts remember the hardware addresses of each other.
• Part of the ARP protocol specifies that the receiving host should also remember the IP and hardware addresses of the sending host.
ARP conversation
RARP conversation
Services provided by IP
• Connectionless Delivery (each datagram is treated individually).
• Unreliable (delivery is not guaranteed).
• Fragmentation / Reassembly (based on hardware MTU).
• Routing.
• Error detection.
IP Datagram
IP Datagram Fragmentation
• Each fragment (packet) has the same structure as the IP datagram.
• IP specifies that datagram reassembly is done only at the destination (not on a hop-by-hop basis).
• If any of the fragments are lost - the entire datagram is discarded (and an ICMP message is sent to the sender).
IP Flow Control & Error Detection
• If packets arrive too fast - the receiver discards excessive packets and sends an ICMP message to the sender (SOURCE QUENCH).
• If an error is found (header checksum problem) the packet is discarded and an ICMP message is sent to the sender.
ICMP
Internet Control Message Protocol
• ICMP is a protocol used for exchanging control messages.
• ICMP uses IP to deliver messages.
• ICMP messages are usually generated and processed by the IP software, not the user process.
ICMP Message Types
• Echo Request
• Echo Response
• Destination Unreachable
• Redirect
• Time Exceeded
• Redirect (route change)
• there are more ...
IP/BYE-BYE
• IP/BYE-BYE is a lecture protocol used to signal the class that we have just finished our discussion of IP - the network layer of TCP/IP.
• The appropriate response to an IP/BYE-BYE request is immediate applause, although simply opening your eyes is enough (known as a WAKEUP response).
Transport Layer & TCP/IP
Q: We know that IP is the network layer - so TCP must be the transport layer, right ?
A: No… well, almost.

TCP is only part of the TCP/IP transport layer - the other part is UDP (User Datagram Protocol).

UDP User Datagram Protocol
• UDP is a transport protocol
• communication between processes

• UDP uses IP to deliver datagrams to the right host.
• UDP uses ports to provide communication services to individual processes.
Ports
• TCP/IP uses an abstract destination point called a protocol port.
• Ports are identified by a positive integer.
• Operating systems provide some mechanism that processes use to specify a port.
Ports
UDP
• Datagram Delivery
• Connectionless
• Unreliable
• Minimal
TCP
Transmission Control Protocol
• TCP is an alternative transport layer protocol supported by TCP/IP.
• TCP provides:
• Connection-oriented
• Reliable
• Full-duplex
• Byte-Stream
Connection-Oriented
• Connection oriented means that a virtual connection is established before any user data is transferred.
• If the connection cannot be established - the user program is notified (finds out).
• If the connection is ever interrupted - the user program(s) is finds out there is a problem.
Reliable
• Reliable means that every transmission of data is acknowledged by the receiver.
• If the sender does not receive acknowledgement within a specified amount of time, the sender retransmits the data.
Byte Stream
• Stream means that the connection is treated as a stream of bytes.

• The user application does not need to package data in individual datagrams (as with UDP).
Buffering
• TCP is responsible for buffering data and determining when it is time to send a datagram.

• It is possible for an application to tell TCP to send the data it has buffered without waiting for a buffer to fill up.
Full Duplex
• TCP provides transfer in both directions (over a single virtual connection).

• To the application program these appear as 2 unrelated data streams, although TCP can piggyback control and data communication by providing control information (such as an ACK) along with user data.
TCP Ports
• Interprocess communication via TCP is achieved with the use of ports (just like UDP).

• UDP ports have no relation to TCP ports (different name spaces).
TCP Segments
• The chunk of data that TCP asks IP to deliver is called a TCP segment.

• Each segment contains:
• data bytes from the byte stream
• control information that identifies the data bytes
TCP Segment Format
Addressing in TCP/IP
• Each TCP/IP address includes:
• Internet Address
• Protocol (UDP or TCP)
• Port Number
TCP vs. UDP
Q: Which protocol is better ?
A: It depends on the application.

TCP provides a connection-oriented, reliable, byte stream service (lots of overhead).

UDP offers minimal datagram delivery service (as little overhead as possible).
TCP/IP Summary
• IP: network layer protocol
• unreliable datagram delivery between hosts.
• UDP: transport layer protocol
• unreliable datagram delivery between processes.
• TCP: transport layer protocol
• reliable, byte-stream delivery between processes.