Recall from Chapter 1, an internetwork is an arbitrary collection of networks interconnected by some
kind of host-to-host packet delivery service.
The network technologies which are joined to the internetwork may be different: for example, FDDI, point-to-point links, and Ethernet.
There also may involve more hosts than could be successfully dealt with by an extended LAN.
The devices that are used to connect different networks are called routers.
The Internet Protocol developed by Kahn-Cerf (1974) is the main protocol used in today's internetworks.
There are other internetwork protocol that have been used to a greater or lesser extent. For example, Novell's IPX, Xerox- XNS, but IP is the dominant protocol.
Service Model
A service model is the set of services we would like a given layer to provide.
For IP, this consists of an addressing scheme that can be used to identify all hosts in the internetwork, and a datagram model (connectionless model) of data delivery.
The service model is best-effort because it makes no guarantees of packet delivery.
Quiz
Which of the following is true?
802.1Q spec is backward compatible with non-VLAN aware bridges.
A CS-PDU is 53 bytes long.
When a crossbar is used as a switch fabric one typically adds to the header of a frame
information to allow the frame to route itself through the fabric.
IP Packet Format
Version - is used to specify the version of IP. Currently, 4 is the most common (IPv4). There is also IPv6.
HLen - specifies the length of the header in 32-bit words. When there are no options, there the header is 5 words longs.
Length - is the length of the datagram in bytes. So the maximum is 65535.
TTL - Time-to-Live is the number of seconds the packet still has to live. Nowadays, it pretty much counts the number of hops. The default start value is 64.
Protocol - identifies the higher level protocol that is being used with IP. 6 = TCP, 17 = UDP
Checksum - adds the header as a sequence of 16 bit words using 1's complement, and takes the 1's complement of the result
Source and Destination address - say where the packet came from and where it is going.
Fragmentation and Reassembly
Ethernet can accept packets up to 1500 bytes long. FDDI can accept packets up to 4500 bytes long.
As IP packets can be much longer, we need some way to split packets so they can fit for a given network type.
To do this a maximum transmission unit (MTU) is defined for each kind of network. This is the largest IP datagram that it can carry in a frame.
Fragmentation of IP frames is then done in a router when it sees IP datagrams which exceed the MTU for the network the datagram is to be sent on.
The resulting fragment will still be in the format of a legal IP datagram.
A receiving router needs to reassemble fragmented datagrams into a complete IP packet.
The Ident field is used to identify all the fragments of a given packet. All fragments of the same packets have the same Ident field.
The DF Flag is used to indicate whether datagram fit all in one fragment or is in more than one fragment.
The MF Flag is used to indicate whether there are more fragments for a given datagram to follow.
Fragmentation always occurs on 8 byte boundaries. The Offset is used to say which such 8 byte boundary in the whole datagram does the data in the given MTU begin with.
So Ident tells us all members of a given datagram, offset allows us to reassemble the datagrams in order.
Global Addresses
Ethernet address are flat -- that is they don't say anything about the network structure.
IP addresses are hierarchical -- they are made of several parts which do say something about the structure
of the internetwork.
For routers each interface might have its own IP address.
Originally, IP networks were divided into three classes: A, B, and C.
If the higher-order bit of the address was 0 it was a class A network, 7 bits would be used for the network identifier and such a network could have up to 224-2 (about 16 million) hosts.
If the high-order two bits were 10, the network was a class B Network, 14 bits would be used for the network identifier and such a
network could have up to 65,535 hosts.
Lastly, if the high-order three bits were 110, 21 bits were used to specify the network and the remaining 8 bits could
be used to say the host.
This original approach turned out to be somewhat inflexible and something called classless inter-domain routing (CIDR) took its place.
We will describe this next week.
IP addresses are typically written as four 8-bit decimal numbers with periods in between them.
Datagram Forwarding and IP
We begin today by looking at how IP packets are forwarded in a network.
In order to understand this it is useful to recall the following things about IP datagrams:
Every IP packet contains the IP address of the destination host.
The "network part" of an IP address uniquely identifies a single physical network that is part of a larger Internet.
All hosts and routers that share the same network part of their address are connected to the same physical network and thus can communicate with each other by sending frames over that network.
The Internet is connected. Further every physical network that is part of the Internet has at least one router that is connected to at least one other physical network; this router can exchange packets with hosts or routers on either network.
To forward a packet a router looks at the network address of the destination and compares it to the network addresses of its interfaces. If it finds a match then it forwards the packet over that interface. Otherwise, it checks if the network address exists in its forwarding tables. If yes, then it forwards according to the table entry. If not, it sends the packet to a default router.
For a host with only one interface and only a default router in its forward table this reduces to: If the network number of the destination is my network number then deliver the packet directly; otherwise deliver the packet to the default router.
Notice that IP is allowing us to be more scalable because we now have a two level hierarchy to get a packet to its destination: we route to the correct network and then use switches in that network to get to the correct host.
Address Translation (ARP)
So far we have explained how IP packets are sent to the correct network.
We still need to explain how they get to the correct host.
The IP address of the destination host is typically different from the kinds of addresses such as Ethernet addresses that
the particular destination network interface cards understand.
To do this a protocol called ARP (Address Resolution Protocol) is used by each host on a network to map between IP addresses and link-level addresses.
The set of mappings held by a given host is called an ARP table or ARP cache.
Since mappings may change over time, entries in these tables have timers associated with them and get timed-out and removed periodically (on the order of 15 minutes).
ARP relies on the fact that many link-layer protocols support broadcast.
If a host wants to send an IP datagram to a host that it knows is on the same network, it first checks for an entry in its table, if it finds an entry it uses it.
Otherwise, the host broadcasts an ARP query on the network.
Each host on the network receives the query and checks to see if it matches its IP address. If it does match, it sends a response message that contains its link-layer address back to the originator of the query. The originator can then add the appropriate entry to its table, and also send the IP datagram.
It should be noted that when a host broadcasts a query message, each other host on the network can learn the sender's link-layer and IP addresses and if they don't have these in their tables they can add them.
ATMARP
ATM networks are often used as part of IP internetworks.
So they too must provide a form of ARP; however, they do not support broadcast. (Remember they are circuit-switched networks.)
ATMARP is the protocol for this and forms part of the so-called classical IP over ATM model.
The basic idea is that ATM networks are split into subnets called logical IP subnets (LISs).
All hosts in the same subnet have the same IP network number.
Two hosts in the same LIS can directly communicate with other; but ones in different LISs cannot.
An ARP server in each LIS is used to maintain a table of mappings between ATM addresses and IP addresses.
Each node in the LIS must be configured to know the ATM address of this ARP server.
When a host boots it is supposed to send its IP address, ATM address information to the server.
If a given host wants to send an IP packet to another host in the same LIS, it makes a VC to the ARP server, gets the
entry, and uses this to make a VC to the desired host.
Host Configuration
Ethernet addresses are globally unique but do not reflect the internetwork structure.
IP addresses need to be globally unique and reflect this structure. So they can't be pre-configured
into a host because a manufacturer can't in general know where you want to connect a host into the Internet.
For this reason, IP addresses need to be reconfigurable. Most OS's provide a means to manually configure the IP information.
If every machine had to be manually configured, this would be a lot of work, and error prone.
DHCP
To solve the problem of manually configured IP addresses, an automated method is typically used. DHCP (dynamic host configuration protocol) is the most common such approach.
DHCP relies on the existence of a DHCP server that is responsible for providing configuration information to hosts.
To contact a DHCP server, a newly booted or attached machine sends a DHCPDISCOVER message (this uses UDP) to 255.255.255.255. This is
a broadcast IP address.
On each network (making use of DHCP) there is at least one relay agent or DHCP server. Relay agents forward DHCPDISCOVER message to a designated DHCP server.
The DHCPDISCOVER message has the host's Ethernet address. The DHCP server responds by sending the message back to the host with its yiaddr (your internet address) field filled. The server adds the Ethernet IP address pair to its table.
In DHCP, IP addresses are leased for a given period of time, after which the lease expires and the server can remove the entry from its tables. Hosts must periodically either renew their leases or get new addresses.
