1. Review of TCP/IP

2. TCP/IP Four layer Architecture Developed in 1960’s Open System
Not just one protocol, whole family.
Many programming interfaces available.
Standardised protocol set.

3. IP Addressing Scheme
Need capability of mapping addresses of one type onto another.
LAN address, Network Point of Attachment NPA, must be mapped onto an IP address.
NPA formats differ from one LAN standard to another.
IP addresses are homogenous within single IP version.

4. IP Address Format 24 bits 7 bits netid hostid Class A 16 bits 14 bits
netid
hostid
Class A
16 bits
14 bits
Class B 10
netid
hostid
21 bits
8 bits
Class C
110
netid
hostid
28 bits
Class D
1110
Multicast group ID

5. IP Address Format (cont.)
Different size networks may use different address classes, defined by the first few bits in the address. 0 for Class A, 10 for Class B, 110 for Class C, etc. etc.
Networks with large numbers of hosts may use Class A, while Class C may have many subnets with a small number of attached hosts.

6. IP Address Notation
A decimal dot notation is used to break down the IP address.
Example
gives the address aka boole !
Note that this is a Class B address (first zero in second position) and the subnet is defined with 14 bits, the host address with 16 bits.

7. IP Allocations
A central authority has responsibility for allocation of IP addresses. They are the network Information center, or NIC.

8. Specail IP Addresses Class D addresses are for multicasting.
Class E are experimental
Private blocks include
– ( /8)
– ( /12)
– ( /16)
Loopback address

9. Subnetting
Subnetting allows for the creation of multiple logical networks within a single Class A, B or C network
Instead of using 16 bits for the hosts, divide the host space up into 2, a subnet and a host
If you have a Class B network, you can connect up to 64 thousand hosts. Think of DCU. Need to break up network into EE, CA, Communications, etc., so we subnet the network
Subnet masks
Class A =
Class B =
Class C =

10. Subnetting a Class B network
512 networks, 126 hosts /25 10
Network
Subnet
Network
Host
254 networks, 254 hosts /24 10
Network
Subnet
Network
Subnet
Host
128 networks, 510 hosts /23 10
Network
Subnet
Network
Subnet
Host
64 networks, 1,022 hosts /22 10
Network
Subnet
Network
Subnet
Host 10
Network
Subnet
Network
Subnet
Host
32 networks, 2,046 hosts /21

11. Subnetting
Normally when a router receives a packet it looks at the IP address and decides if it is local or has to sent elsewhere. Entries look like (network, 0) and (this-network, host). The routing table has entries for local packets as well as distant packets. A router only needs to know about its local hosts, some other networks and where to send all other packets
With subnetting an extra entry is added to the routing table stating (this-network, subnet, 0) and (this-network, this-subnet, host)
This way a router knows about all of its own hosts and how to get to the other subnets

12. Subnet Mask
Router has a subnet mask telling it the split between (subnet) networks and hosts
Subnetting is not visible outside the network
Boolean AND to remove host part

13. Subnetting
To recap, subnetting divides an organisations single class A, B or C network into multiple logical networks by dividing the original host identifier string into two, with the first string representing the subnet and the second representing the hosts
Routers use a subnet mask to determine if a packet is to be routed to the current network, another network in the subnet or a distant network

14. TCP/IP Encapsulation Application TCP IP 802.3 user Data Appl Hdr user
header
Application Data
TCP IP
header
TCP
header
Application Data IP
Ethernet
header IP
header
TCP
header
Application Data
Ethernet
trailer
802.3

15. TCP Segment Header 16-bit source port number
16-bit destination port number
32-bit sequence number
32-bit acknowledgement number
4bit hdr
length
reserved u r g A C K P S H R S T S Y N F I N
16-bit window size
16-bit TCP checksum
16-bit urgent pointer
Options (if any)
Data (if any)

16. TCP Header Description
Source Port and Destination Port identify transport end-points of connection.
Sequence Number and Acknowledgement Number perform usual functions, Ack numbers next byte expected.
TCP Header Length indicates number of 32 bit words in header. Length varies because of options.
Not used. No bug fixes required !

17. Six one bit flags…
URGent pointer in use, used for indicating interrupts and offset from seq no. to urgent data.
ACK bit used to indicate piggybacked acknowledgement.
PSH requests that receiver does not buffer but to deliver.
RST is reset connection, means problems !
SYN used in conjunction with ACK to request connection.
FIN release connection

18. Window size used for variable-sized sliding window
Window size used for variable-sized sliding window. Size of zero indicates a choke packet.
Checksum checks header.
Options field for things like specification of maximum TCP payload. Negotiated at startup lowest bid wins.
A selective repeat instead of go-back-n sliding window protocol may be specified as an option.

19. TCP Addressing
TCP uses notion of Port Number to access transport endpoint on a single host. Many Ports may be in use simultaneously.
Combination of IP address and port number uniquely identifies a port for process running on a particular machine.
Process may even have several ports open.

20. TCP Services
Provides connection-oriented, reliable, byte stream service.
Segments passed to IP for routing, timer attached for each segment.
Sliding window protocol utilised with go-back-n or selective-repeat for retransmission.
All TCP segments acknowledged.

21. TCP segments may arrive out of order, sliding window will sort order.
TCP segments may be duplicated, duplicated are discarded.
TCP provides flow control, no process\host will be swamped, helps avoid congestion.
TCP utilised by many internet applications such as Telnet, Rlogin, FTP, , WWW Browsers.

22. What programmers need to know
TCP – The Guts
What programmers need to know

23. Three Way Handshake Socket, Bind, Listen Socket Accept(blocks)
Connect(blocks)
(active open)
SYN J
SYN K, ack J+1
Connect returns
Ack K+1
Accept returns
Read(blocks

24. Server prepares connection
socket, bind, listen. This is a passive open.
Client issues active open by calling connect.
Issues a SYN segment with sequence no.
Contains IP header, TCP header and possible TCP options (next slide)
Server acks clients SYN with its own SYN with initial sequence no that server will send. The SYN and ACK are sent is the same segment.
Client acknowledges the servers SYN
3 packets are sent (minimum) so called 3-way-handshake.

25. Common TCP Options in SYN
MSS: maximum segment size (Stevens Ch 7.9)
Window Scale Option: max window (16 bit size). Window may be scaled (left shifted) by 0-14 bits giving amx window size of x 214. Only used if both sender and receiver agree.
Timestamp option: used on hi-speed connections to prevent corruption due to reappearing packets, negociated similarely to above.
Latter two called RFC 1332 options, or “long fat pipe options”

26. TCP Connection Termination
If application calls close forst, this is an active close.
Sends FIN segment, meaning finished sending data.
Server performs passive close.
Clients FIN is ack’ed and sent to application as EOF, after any queued data to receive.
When application receives its EOF, it will close its socket. TCP sends FIN.
The server on receiving final FIN acks that FIN.

27. FIN M ack M+! FIN N ack N+1 Close (active close) (passive close)
read returns 0
close
ack M+!
FIN N
ack N+1

28. TCP SDT Normal client transitions Normal Server transitions
appl: Application issues operation
recv: segment received
send: what is sent for this transition
The netstat application uses the state names from this diagram, try it out.

29. starting point closed listen active open SYN_RCVD SYN_SENT ESTABLISHED
appl:passive open
send<nothing>
Appl: active open
Send SYN
listen
recv: SYN; send: SYN, ACK
active open
recv: RST
recv: SYN
send: SYN, ACK
simultaneous open
SYN_RCVD
SYN_SENT
recv:ACK
send<nothing>
recv:SYN, ACK
send: ACK
ESTABLISHED
data
Transfer
state
CLOSE_WAIT
recv: FIN
send: ACK
appl: close
send: FIN
appl: close
send: FIN
recv: ACK
send: <nothing>
LAST_ACK

30. CLOSING FIN_WAIT_1 TIME_WAIT FIN_WAIT_2 appl: close send: FIN
simultaneous close
recv:FIN
send: ACK
FIN_WAIT_1
CLOSING
recv: FIN, ACK
send: ACK
recv: ACK
send: <nothing>
recv: ACK
send: <nothing>
recv: FIN
send: ACK
TIME_WAIT
FIN_WAIT_2
2MSL timeout

31. 11 states defined. Rules of TCP apply…
If application performs active open in CLOSED state, TCP sends SYN and new state is SYN_SENT.
If TCP next receives a SYN with an ACK, it sends an ACK and the new stste is ESTABLISHED
Two arrows leading from ESTABLISHED deal with termination.
If application calls close before receiving eof (active close), transits to FIN_WAIT_1
If application receives FIN while ESTABLISHED (passive close), transits to CLOSE_WAIT

32. TCP Connection & The Packets
A complele TCP connection involves many packet exchanges.
Connection establishment
Data transfer
Connection termination
TCP states are also shown as client and server enter them.

33. Client Server socket, bind, listen LISTEN(passive open) accept(blocks)
SYN_RCVD
ESTABLISHED
accept returns
read(blocks)
read returns
<server process request>
write
CLOSE_WAIT(passive close)
read returns 0
close
LAST_ACK
CLOSED
SYN J, mss=1460
Socket
Connect(blocks)
(active open) SYN_SENT
ESTABLISHED
Connection returns
<client forms request>
Write
Read(blocks)
Read returns
Close
(active close) FIN_WAIT_1
FIN_WAIT_2
TIME_WAIT
SYN K, ack J+1, mss=1024
ack K+1
Data (request)
Data reply
Ack of request
ack of reply
FIN M
ack M+!
FIN N
ack N+1

34. Client announces MSS 1460, typical for Ethernet, Ok if different in each direction.
Once connection established, clients forms request for server.
Server processes request and replies with piggybacked ack.
Termination by client (active close) enters TIME_WAIT state, 2MSL (Maximum Segment Lifetime) to deal with lost or wandering IP packets.