• IP Service
• IP Addressing
• Datagram Format
• Routing (Forwarding)
• Subnetting
• Fragmentation
© Jörg Liebeherr (modified by M. Veeraraghavan)
1
Orientation
• IP (Internet Protocol) is a Network Layer Protocol.
TCP
UDP
Transport
Layer
ICMP
IP
IGMP
Network
Layer
ARP
Network
Access
RARP
Link Layer
Media
• IP’s current version is Version 4 (IPv4). It is specified in RFC
891.
© Jörg Liebeherr (modified by M. Veeraraghavan)
2
Orientation
• IP layer processing occurs at both routers and hosts:
Application
Application protocol
Application
TCP
TCP protocol
TCP
IP
Network
Access
Host
IP
IP protocol
Data
Link
Network
Access
Network
Access
Router
© Jörg Liebeherr (modified by M. Veeraraghavan)
IP
IP protocol
Data
Link
Network
Access
IP protocol
Network
Access
Router
Data
Link
IP
Network
Access
Host
3
IP Service
• IP provides an unreliable and connectionless service (“datagram service”).
Unreliable: IP does not guarantee that a transmitted
packet will be delivered.
Connectionless: Each packet (“datagram”) is
handled independently. IP is not aware
that packets between hosts may be sent in
a logical sequence.
• Consequences of an unreliable, connectionless service
– Lost packets
– Packets delivered out-of-sequence
– Can have duplicate packets because of TCP layer resends*
• e.g. if packet is delivered but ACK is lost, TCP sender will resend
packet buffered at the end host.
*: The audio file says duplicate packets could occur because of router-buffered packets;
this is not commonplace in today’s networks because common link-layer protocols do not
have retransmissions, but if they did, this could be another cause of duplicate packets;
for example, 802.11 wireless link layer has retransmissions.
© Jörg Liebeherr (modified by M. Veeraraghavan)
4
IP Service
• IP offers a best effort * service;
i.e., IP does not make performance guarantees on:
• the time until a packet is delivered
• the packet loss rate
• the throughput of traffic between two hosts
• etc.
• Performance guarantees are also called Quality-of-Service
or QoS guarantees.
• IP provides limited QoS support. This is done via the Type-ofService (TOS) field.
* “Best effort” may mean different things to different people.
© Jörg Liebeherr (modified by M. Veeraraghavan)
5
IP Service
• IP supports the following services:
• one-to-one
(unicast)
• one-to-many
(multicast)
• one-to-all
(broadcast)
Unicast
Broadcast
Multicast
• IP multicast actually supports a many-to-many service.
• IP multicast requires support of other protocols (IGMP,
multicast routing).
© Jörg Liebeherr (modified by M. Veeraraghavan)
6
Internet Addresses
• Each network interface on the Internet as a unique global
address, called the IP address.
• An IP address:
- is 32 bits long.
- encodes a network number and a host number
• IP addresses are written in a dotted decimal notation :
- 128.238.42.112
10000000
11101110
00101010
01110000
© Jörg Liebeherr (modified by M. Veeraraghavan)
means
in 1st Byte
in 2nd Byte
in 3rd Byte
in 4th Byte
7
Internet Address Classes
• IP distinguishes 5 classes of addresses.
Class A
Class B
Class C
Class D
0
network id
host
7 bits
24 bits
1 0
network id
host
14 bits
16 bits
1 1 0
1 1 1 0
network id
host
21 bits
8 bits
multicast group id
28 bits
Class E
1 1 1 1 0
(reserved for future use)
27 bits
© Jörg Liebeherr (modified by M. Veeraraghavan)
8
IP Addresses
• The end points of each range are not allowed because all
zeros and all ones are disallowed for netid, subnetid and
hostid (see pg. 42 of text and RFC791).
Class From
To
A
0.0.0.0
B
128.0.0.0 191.255.255.255
C
192.0.0.0 223.255.255.255
D
224.0.0.0 239.255.255.255
E
240.0.0.0 247.255.255.255
© Jörg Liebeherr (modified by M. Veeraraghavan)
127.255.255.255
9
Trade-off of Address Classes
• There are a total of 232 = 4,294,967,296 IP addresses
Class A:
7 bits for netid 
only 128 Class A networks
each net can have about 16 million (224-2) hosts.
Class B:
14 bits for netid 
about 16,000 networks
About 65,000 (216 -2) hosts per network
Class C:
21 bits for netid 
about 2 million networks
Only 254 hosts per network
© Jörg Liebeherr (modified by M. Veeraraghavan)
10
Special case IP addresses (not complete)
• 0: all zeros (means “this host” or “this net”); -1: all ones (this is reserved for
broadcasts); netid, subnetid, hostid: anything else
IP address
netID
subnetID
Can appear as
Description
hostID
Source? Destination?
0
0
OK
never
This host on this net
0
hostid
OK
never
Specified host on this net
127
anything OK
OK
Loopback address
-1
-1
never
OK
Limited broadcast
netid
-1
never
OK
Net-directed broadcast to
netid
Subnet-directed broadcast
to netid, subnetid
All-subnets-directed
broadcast to netid
netid
subnetid
-1
never
OK
netid
-1
-1
never
OK
© Jörg Liebeherr (modified by M. Veeraraghavan)
11
IP Datagram Format
• 20 bytes  Header Size  24 * 32 bit-words = 60 bytes
• 20 bytes  Total Length  216 bytes = 65536 bytes
version
(4 bits)
header
length
Type of Service/TOS
(8 bits)
>= five 32-bit words
Identification (16 bits)
TTL Time-to-Live
(8 bits)
Total Length (in bytes)
(16 bits)
flags
(3 bits)
Protocol
(8 bits)
Fragment Offset (13 bits)
Header Checksum (16 bits)
Source IP address (32 bits)
Destination IP address (32 bits)
Options (if any, <=40 bytes)
DATA
0
31
32-bit word
© Jörg Liebeherr (modified by M. Veeraraghavan)
12
IP Datagram Transmission
• Question: If you measure an IP datagram on the network, in
which order are the fields transmitted?
• Answer:
• Transmission is row by row
• For each row:
1. Transmit bits 0-7
2. Transmit bits 8-15
3. Transmit bits 16-23
4. Transmit bits 24-31
• This is called network byte order or big endian byte
ordering.
• Note: some computers store 32-bit words in little endian format. Which
ones?
© Jörg Liebeherr (modified by M. Veeraraghavan)
13
Fields of the IP Header
• Version: current version is 4, next version will be 6.
• Header length (4 bits): Number of 32-bit words in the IP
header
• Type of Service (TOS): contains 4 TOS bits, where each bit
indicates a desired service.
• minimize delay
• maximize throughput
• maximize reliability
• minimize cost
– Only one bit can be set! Not supported in all applications.
© Jörg Liebeherr (modified by M. Veeraraghavan)
14
IP Header Fields Contd.
• Total Length: Number of bytes in the IP datagram
(header+payload)
• Identification: unique identification of a datagram from a
host. Incremented whenever a datagram is transmitted.
• Flags and Fragment Offset: Associated with fragmentation
• Time To Live (TTL): specifies longest path before datagram
is dropped.
– Used to prevent infinite looping of packets
– TTL field is set at sending host and is decremented by 1 at each router
– If a router gets a datagram whose TTL is either 0 or 1, the router will
drop the packet.
– If a destination host gets a datagram whose TTL is 1, the host will
deliver the datagram to the higher layer.
© Jörg Liebeherr (modified by M. Veeraraghavan)
15
Fields of the IP Header
• Protocol: Specifies the higher-layer protocol.
Used for demultiplexing to higher layers.
Application
TCP
UDP
ICMP
IGMP
IP
• Header checksum: verifies correctness of header.
© Jörg Liebeherr (modified by M. Veeraraghavan)
16
Fields of the IP Header
• Source and Destination Addresses: identify the interfaces
on the sending and receiving hosts
• Options:
• Security restrictions
• Record Route: each router that processes the packet adds its IP
address to the header.
• Timestamp: each router that processes the packet adds its IP
address and time to the header.
• (loose) Source Routing: specifies a list of routers that must be
traversed.
• (strict) Source Routing: specifies a list of the only routers that
can be traversed.
• Padding: ensures that header ends on a 4-byte boundary
© Jörg Liebeherr (modified by M. Veeraraghavan)
17
Routing
• The routing functions at the network layer must be able to
deliver a packet from a source host to a destination host.
• There are two parts to the routing problem:
1. How to pass a packet from an input interface to
the output interface of a router (“packet forwarding”) ?
2. How to create routing tables?
• In this lecture, we only discuss the packet forwarding part.
The other function is discussed in Lecture 6.
© Jörg Liebeherr (modified by M. Veeraraghavan)
18
Major Tenets for Datagram Forwarding
• Every IP datagram contains the IP address of an interface of
a destination host.
• The network id of an IP address uniquely identifies a
single physical network that is part of the larger Internet.
• All hosts and routers that have the same network id are
connected to the same physical network and can directly
communicate by sending MAC layer frames.
• Every physical network of the Internet has at least one router,
which is also connected to at least one other physical
network.
© Jörg Liebeherr (modified by M. Veeraraghavan)
19
IP Forwarding
• A packet is typically forwarded to a large number of routers
before reaching the destination host.
• IP forwarding is done on a hop-by-hop basis,
i.e., no one knows the complete route. The goal of forwarding
is to bring the IP datagram closer to the destination.
© Jörg Liebeherr (modified by M. Veeraraghavan)
20
IP Forwarding
• IP forwarding is performed by both hosts and routers.
• The difference between IP forwarding in a host and in a router is that a
host’s IP module does not forward packets received on an interface to
another interface (if it does then it is behaving as a router). In a host, IP
forwarding is from the higher layers to an interface or vice versa
• Both routers and hosts have a routing table. Routing table entries look like
this and is looked up for each datagram:
Destination IP address of
Flags
IP address Next-Hop-Router
Specification
of an interface
Whole IP
Address of router
address or
interface or directly
network ID of connected network
IP address
Interface to which
the datagram is
passed
© Jörg Liebeherr (modified by M. Veeraraghavan)
21
Flags
• H: Whether the destination address is the network ID of an IP
address (H=0) or the complete IP address of a host interface
(H=1).
• G: Whether the next-hop router field is really a next-hop router
(G=1) or a directly connected interface (G=0)
© Jörg Liebeherr (modified by M. Veeraraghavan)
22
Forwarding Functions at Host
• Forwarding an IP datagram at a host:
DestNet = Network id in destination IP address
If (DestNet == my Network id)
Deliver datagram to destination directly;
else
Send datagram to default router;
• Reception of an IP datagram at a host:
Datagram is received on an interface
If (Destination IP address == (my IP address ||
broadcast address))
Deliver datagram to higher layer;
else
Discard the datagram;
© Jörg Liebeherr (modified by M. Veeraraghavan)
23
Forwarding Functions at Router
• Reception of an IP datagram at a router:
Packet is received on an interface
If (Destination IP address != my IP address)
Perform datagram forwarding;
• Forwarding an IP datagram at a router:
DestNet = Network id in destination IP address
DestIP = destination IP address
If (DestIP == Destination IP address in my routing
table)
Deliver datagram to the next-hop-router in the table;
elseif (DestNet == Network id in my routing table)
Deliver datagram to destination of the interface;
else
deliver packet to default router;
© Jörg Liebeherr (modified by M. Veeraraghavan)
24
Example (from Textbook)
ftp.uu.net
192.48.96.9
140.252.1.0
140.252.13.0
140.252.1.183
140.252.104.1
140.252.1.4
Gateway
netb
modem modem
140.252.1.29
140.252.13.33
sun
140.252.13.35
bsdi
• Scenario 1: bsdi sends a datagram to sun
• Scenario 2: bsdi sends a datagram to ftp.uu.net
© Jörg Liebeherr (modified by M. Veeraraghavan)
25
Scenario 1
140.252.13.0
140.252.13.33
sun
140.252.13.35
bsdi
• Bsdi looks up its routing table and finds sun as being located on the same
network.
• How ? The network address of sun and bsdi match.
•Bsdi issues an ARP request for 140.252.13.33
•Sun replies with the MAC address for its interface
•Bsdi can then send the IP datagram encapsulated in an ethernet frame
© Jörg Liebeherr (modified by M. Veeraraghavan)
26
Scenario 2
ftp.uu.net
192.48.96.9
140.252.1.0
140.252.13.0
140.252.1.183
140.252.104.1
140.252.1.4
Gateway
modem modem
140.252.1.29
netb
140.252.13.33
sun
140.252.13.35
bsdi
• bsdi sends a datagram to ftp.uu.net.
1. bsdi sees that its network ID differs from that of the destination address
2. So it sends the datagram to its default router (sun) (by using the MAC
address of the 140.252.13.33 interface on sun - obtained by ARP)
© Jörg Liebeherr (modified by M. Veeraraghavan)
27
Scenario 2, Part II
ftp.uu.net
192.48.96.9
140.252.1.0
140.252.13.0
140.252.1.183
140.252.104.1
140.252.1.4
Gateway
netb
modem modem
140.252.1.29
140.252.13.33
sun
140.252.13.35
bsdi
3. sun (is configured as router) forwards the packet to its default router
(which is set to netb).
4. netb forwards the datagram to its default next-hop-router, set to
Gateway (140.252.1.4).
5. Gateway forwards the packet to somewhere else (not known).
© Jörg Liebeherr (modified by M. Veeraraghavan)
28
Subnetting
• The address classes makes Class A and B addresses very
attractive and Class C addresses not attractive:
- The number of network id’s relates to the size of the
routing table in IP routers.
- The number of hosts on a network is limited to 255.
- Organization must contact IANA for each new
LAN.
- Management of a large number of Class C
addresses is cumbersome.
• But: There are only few Class A and B addresses!
• Solution:
© Jörg Liebeherr (modified by M. Veeraraghavan)
Subnetting
29
Subnetting
• Goal: Reduce the need for a large number of network ids.
• Basic Idea:
- Split the host id portion of an IP address into a
subnet id and a host id.
- Assign one subnet id to each physical
network.
• Then:
• Subnets can be freely assigned and be used for many
physical networks.
• Distant routers need not be aware of subnet id’s.
© Jörg Liebeherr (modified by M. Veeraraghavan)
30
Subnet Masks
• Routers and hosts use a subnet mask to separate the
subnetwork id from the host id.
Class B
1 0
network id
host
14 bits
with
subnetting
1 0
16 bits
network id
14 bits
Subnet
mask
subnet id
host
8 bits
8 bits
111111111111111111111111
00000000
24 bits
8 bits
(255.255.255.0)
Question: In the IP address 152.164.9.10 with subnet mask 255.255.255.128, what class
does this IP address belong to and what are the 14 bits netid, subnet id and hostid? How
many bits are in the subnet id and in the hostid?
Answer: Class B; 14 bits netid = 24.164, subnet id (9 bits) = 9.0 hostid (7 bits) = 10
© Jörg Liebeherr (modified by M. Veeraraghavan)
31
Subnetting: Host Processing
• Each host has to know the IP address and the subnet mask of its interface
• For each IP datagram received from the application, the IP layer at the
host peforms:
if ((Destination IP address & Subnet Mask) ==
(My IP address & Subnet Mask))
Deliver datagram directly (perform ARP if shared medium LAN);
else
Forward datagram to default router;
© Jörg Liebeherr (modified by M. Veeraraghavan)
32
Subnetting: Router Processing
• A router has a table:
< SubnetNumber, SubnetMask, NextHop>
» SubnetNumber is the <network id, subnet id>
» SubnetMask is the subnet mask
» NextHop is the IP address of a router.
• The router perfoms the following algorithm:
– Reception of an IP datagram at a router:
Packet is received on an interface whose IP address
is IP1.
dest = destination IP address;
If (dest == IP1)
Deliver datagram to TL/application in router ;
© Jörg Liebeherr (modified by M. Veeraraghavan)
33
Subnetting: Router processing Contd.
Else
for each entry <SubnetNumber, SubnetMask, NextHop>
sub1 = dest & SubnetMask;
if (sub1 == SubnetNumber)
Forward datagram to NextHop;
else
Go to next entry in routing table;
© Jörg Liebeherr (modified by M. Veeraraghavan)
34
Advantages of Subnetting
• Improves efficiency of IP address usage by not consuming an
entire Class B or Class C address for each physical network.
• Reduces routing table sizes. Since external routers do not
know about subnetting, the size of routing tables is reduced.
• Networks become easier to manage.
• Note that there are three levels of aggregation:
• Network
• Subnet
• Host
© Jörg Liebeherr (modified by M. Veeraraghavan)
35
IP Fragmentation
• Host A sends a large IP datagram to Host B.
• Any Problem with that?
FDDI
Ring
Host A
MTUs:
FDDI: 4352
Ethernet
Router
Host B
Ethernet: 1500
• IP router splits the datagram into several fragments
• Fragmentation requires that the data portion of every
fragment except the last be a multiple of 8 bytes.
© Jörg Liebeherr (modified by M. Veeraraghavan)
36
Where is Fragmentation done?
• Fragmentation can be done at the sender or at
intermediate routers.
• The same datagram can be fragmented several times.
• Reassembly of original datagram is only done at
destination hosts.
IP datagram
H
Fragment 2
H2
Fragment 1
H1
Router
© Jörg Liebeherr (modified by M. Veeraraghavan)
37
What’s involved in Fragmentation?
• The following fields in the IP header are involved:
version
(4 bits)
header
length
Type of Service/TOS
flags
Identification
TTL Time-to-Live
(8 bits)
Total Length (in bytes)
Protocol
(8 bits)
Fragment Offset
Header Checksum (16 bits)
......
Identification
Flags
is the same in all fragments.
contains a “more fragments” bit.
(There is also a “don’t fragment bit” that can be set).
Fragment offset
Total length
© Jörg Liebeherr (modified by M. Veeraraghavan)
contains the offset (in 8-byte units) of
current fragment in the original datagram.
is changed to be the size of the fragment.
38
Multiple links
• The ID field stays the same for all fragments of a datagram sent by a
sender to allow for reassembly
• The fragment offset is relative to the datagram sent by the sender.
• Two fragments created on X.25 link (offsets 0, 69)
– 576 – 20 (IP header) = 556; 552 divides by 8 as 69.
• Each is fragmented further on the PPP link
– ID stays the same on all fragments
– Fragment offset on the second set of fragments is relative to the
original (0, 34, 68, 69, 103)
• 296-20=276; 272/8 = 34
IP datagram sent has a
payload of 1000 bytes
© Jörg Liebeherr (modified by M. Veeraraghavan)
X.25
Router (MTU=576)
1
PPP
Router (MTU=296)
2
39
Configuring a Network Interface
• The ifconfig command is used for setting and testing
network interfaces (in Unix systems) that run TCP/IP
• Ifconfig must be run at boot time to initialize the network
interfaces.
• Ifconfig -a plots a report on all interfaces.
: aida:/ ; ifconfig -a
ed1: flags=8863<UP,BROADCAST,NOTRAILERS,RUNNING,SIMPLEX,
MULTICAST> mtu 1500
inet 128.238.42.114 netmask 0xffffff00 broadcast
128.238.42.255
lo0: flags=8049<UP,LOOPBACK,RUNNING,MULTICAST> mtu 16384
inet 127.0.0.1 netmask 0xff000000
© Jörg Liebeherr (modified by M. Veeraraghavan)
40
Configuring a Network Interface
• The netstat command plots (a lot of) network related data
structures in the operating system.
• Netstat -i plots information on the network interfaces.
: aida:/ ; netstat -i
Name Mtu
Network
Address
ed1
1500 128.238.42.114 aida
lo0
16384 your-net
localhost
© Jörg Liebeherr (modified by M. Veeraraghavan)
Ipkts Ierrs
Opkts Oerrs Coll
2394482 35664 346880
223 119591
153
0
153
0
0
41