CS485G Spring 2015 62
(b) Software on hosts and routers agree on a common internet pro-
tocol independent of the capacities of routers and links.
9. IPv4 is a widely used internet protocol.
(a) Uniform format for addresses: 32 bits, typically expressed as 4
octets: 128.163.146.21
(b) Each host and router has at least one IP address; these addresses
are unique across the Internet (exceptions discussed later).
(c) Uniform data-transfer unit: packet, consisting of a header and
a payload.
(d) The header has fields such as packet size (between 20 and 65,535
bytes), source address, destination address.
(e) The payload is composed of arbitrary data.
(f) Packets are embedded in frames for transit across individual
links.
10. Complexities covered in a course in computer networking
(a) Different networks may have different maximum frame sizes,
leading to packets becoming fragmented, so the protocol must
include an ability to put the fragments back together.
(b) Packets may arrive out of order at the destination, so the proto-
col must include an ability to put them back in order.
(c) Packets may be corrupted in transit or otherwise lost, so the
protocol must be able to discover the problem (typically by lis-
tening for acknowledgements) and retransmit packets.
(d) Routers need to know how to forward packets, and they need
to keep up to date with network-topology changes.
(e) The transmission protocol needs to deal with congestion by re-
ducing bandwidth use.
11. TCP/IP protocol family
(a) IP (internet protocol): addressing scheme, host-to-host unreli-
able, unordered delivery.
(b) UDP (unreliable datagram protocol): uses IP for process-to-process
unreliable, unordered delivery.
(c) TCP (transmission control protocol): uses IP for reliable, in-
order process-to-process communication over established con-
nections.
Comentarios a estos manuales