Transcript presentation source

Chapter 14
TCP/IP and Routing
Part #1
Unix System Administration
T.C. P’s and I P’s






What is TCP/IP?
TCP/IP is a networking protocol for
transmitting data between computers
TCP/IP is a family of protocols
TCP/IP is the protocol of the Internet
TCP/IP is the default protocol of UNIX
TCP/IP provides a uniform programming
interface to different types of network
hardware that guarantees interoperability
We Are Family
IP - Internet Protocol which transports
raw data from one machine to another
 ICMP - Internet Control Message
Protocol provides low-level support for
IP such as error messages, routing
assistance and “pings”
 ARP - Address Resolution Protocol
which translates logical address (IP
address) to hardware address (MAC
address)

All in the Family
TCP - Transmission Control Protocol
uses IP to maintain a connectionoriented sessions between machines
that guarantees delivery of data.
 UDP - User Datagram Protocol uses IP
for connectionless sessions between
machine without the guarantee of data
delivery.

So You Want to Date a Model?
The OSI Model by the ISO
 Seven Layers

– Layer
– Layer
– Layer
– Layer
– Layer
– Layer
– Layer
7
6
5
4
3
2
1
-
Application Layer
Presentation Layer
Session Layer
Transport Layer
Network Layer
Data Link Layer
Physical Layer
The Government Thought OSI
Was A Good Idea
The US Government decided all
government computers should use OSI
 OSI had implementation problems

– Relied on concepts that made no sense in
modern networks
– Some specifications were incomplete
– Functionally inferior to existing protocols
– Too many layers which made it slow
Back to Good Ol’ TCP/IP in the
Good Ol’ US of A
Good enough of government work
 TCP/IP has 4 layers

– Layer 4 - Application Layer (think telnet,
ftp, http, etc.)
– Layer 3 - Transport Layer (think TCP/UDP)
– Layer 2 - Network Layer (think IP)
– Layer 1 - Link Layer (think ethernet card)
How Does it All Fit Together
Exhibit A from Page 246 in your book
Pack-et In a Frame
Data travels the network in a packet
which consists of a payload and
headers
 Packets are sometimes referred to as
frames when talking about the low-level
transport, (e.g. Ethernet)
 The low-level transport may limit the
size of a frame/packet. This is called
the MTU - maximum transfer unit

Sample Packet
IP on Your MAC Port?

Three key addresses determine the
source and destination of a packet
– MAC - low-level address of network
hardware. For Ethernet, 6-byte globally
unique address.
– IP - dotted quad-octet assigned to each
network device. Globally unique.
– Port - two-byte number that specifies a
particular service on a machine
Mapping IP to MAC Using ARP

The ARP protocol is used to map IP
address to MAC addresses
– Machine sends ARP request “Who is
159.91.15.220?”
– The machine or another machine on the
same network will respond
“00:40:95:21:CB:38 is 159.91.15.220”
– The MAC address is used to make the final
delivery of the packet to the destination
machine.
Dotted? Quad? Octet?

IP Address
– Octet = 8-bit number (usually we call this a
byte, but old DEC systems used 7-bit
bytes) = values 0-255
– octet.octet.octet.octet is the format of an
IP address, such as 159.91.15.220
– IP addresses have a structure to their
assignment, where as, MAC addresses
have no relationship to each other
IP Address Have Classes Too

There are 5 classes for IP addresses
– Class A - first octet is assigned as the
network address, the remaining 3 octets
are available for network devices
– Class B - first two octets are assigned as
the network address, remaining 2 octets
are available for network devices
– Class C - first three octets are assigned as
the network address, remaining octet is
available for network devices
Going Out With Class

The last two IP classes are special
– Class D - multicast addresses. This is the
rare case where the multiple machines can
have the same IP address
– Class E - experimental
IP Class Summary
Class
1st Octet
Format
A
1-126
N.H.H.H
B
128-191
N.N.H.H
C
192-223
N.N.N.H
D
224-239
-
E
240-254
-
Who’s The Head of the
Classes?




*NOTE* - the book is out of date about this
Today, you typically get your addresses from
your Internet Service Provider
Addresses are ultimately controlled by IANA Internet Assigned Number Authority. IANA
give certain registries around the globe
authority to give out IP addresses.
www.iana.org
ARIN - American Registry for Internet
Numbers handles the Americas. www.arin.net
IP Privately
There are private IP addresses set aside
for internal use. They are not routed if
they get to the Internet.
 These private addresses can be used if
you setup a standalone network or if
you will be using NAT (Network Address
Translation) to translate a private
address to a valid public address before
going to the Internet.

Still, IP Privately

RFC1918
– http://www.isi.edu/in-notes/rfc1918.txt
– Class A - 10.0.0.0 thru 10.255.255.255
– Class B - 172.16.0.0 thru 172.31.255.255
– Class C - 192.168.0.0 thru 192.168.255.255
What Service Do You Use Your
Port For?




A port is the location on a particular machine
that a service is running on.
Think of it as room in a house. The house
has an address (IP), but each room (port)
provides a certain function.
Ports below 1024 are called “well known”
ports. Only root has access to start services
on these ports.
Ports 1024 and above are called “ephemeral”
ports. These ports are open to everyone on
the system.
What’s That In My #2 Port?
Normally, a vendor goes to IANA to get
a port assigned to them for their service
to run on. Each service has a unique
port. E.g. - telnet is 23, http is 80.
 RFC1700 used to keep this list, but is
now out-dated.
 See http://www.iana.org/numbers.html
for the most up-to-date list.

Why, Why, Y?

Why would you use a private address
with NAT when connecting to the
Internet instead of getting public
addresses for each machine?

Why would you use a multicast (class
D) address?
Coming Soon to a Lecture
Near You

Part #2 - Routing and Subnets