computer networks group universität paderborn computer networks chapter 1: a brief kms recap holger...

Computer Networks GroupUniversität Paderborn

Computer Networks Chapter 1: A brief KMS recap

Holger Karl

WS 05/06, v 1.0 Communication Networks - Ch. 1 - A brief KMS recap 2

Goals of this chapter

Briefly repeat the central aspects of communication networks that have been already treated in KMS


Overview

Communication basics Duplexing, multiplexing, medium access Routing, transport Service vs. protocol Reference models


Communication basics

Information: Human interpretation Data: Formalized representation Signal: Representation of data by

characteristic changes of a physical variable

These changes can travel over distance Example: current, voltage, tension, … Can be generated by sender, interpreted

by receiver

Immaterial signals in physical media enable data communication between remote senders and receivers

Signals represent bits

Conventionsfor

representation

Information

Data

Conventionsfor

representation

Abstractworld

Signals

Physicalworld


Signals propagate in medium, store data

Signals traveling in a medium take time to reach destination – delay d

Depends on distance and propagation speed in transmission medium

To represent one or several bits, a signal extending in time is needed – duration of transmission

Determined by rate r and data size

During time d, r*d bits are generated

Stored in the medium

Message Sequence

Charts (MSC)

Start oftransmission

End oftransmission

Delay d

Tim

e

Distance


Basic organization of communication

Duplexing: Given a single pair of communicating peers, duplexing describes rules when each peer is allowed to send to the other one

Using which resource

Mutiplexing: Given several pairs, multiplexing describes when which pair, using which resources, is allowed to communicate

Main resources: Time, frequency (+ some others) Example combinations?


Multiplexing & shared resources

Multiplexing can be viewed as a means to regulate the access to a resource that is shared by multiple users

The switching element/its outgoing line With the switching element as the

controller

Are there other examples of “shared resources”?

Classroom, with “air” as physical medium A shared copper wire, as opposed to direct

connection

Characteristic: a broadcast medium!

Virtually shared, but exclusively controlled!

Shared!

Shared!


Handling many devices: Introducing multiple hops


How to realize multiple hops: Switching

In absence of direct connection between communicating peers, some sort of switching becomes necessary

Option 1: Circuit switching Request a (physical) connection Turn knobs, switches, etc. Use this connection as before – peers are

now directly connected

http://www.wdrcobg.com/switchboard.html


How to realize multiple hops: Switching

Option 2: Packet switching Instead of building and releasing an end-to-end connection for

each communication’s entire length, only Use connections from one hop to another hop Communicate well identified parts of a communication – packets –

between these hop neighbors

Packet 1

Packet 2

Packet 3

To X

Buffer

To YTo Z

To Y To ZTo Z

Pick next hop

To Y To X To Z

Packets need to have addresses to find destination


Forwarding and next hop selection

Recall: A switching element/a router forwards a packet onto the next hop towards its destination

How does a router know which of its neighbors is the best possible one towards a given destination?

What is a “good” neighbor, anyway?

A

Z

?

?

?

?

U

V

W

X

Y

P

M


Provide data for next hop selection

Construct routing tables For each switching

element separately Separate entry for each

destination Contains information

about the (conjectured) shortest distance to a given destination via each neighbor

M P Z

U 2 3 4

V 3 2 3

X 4 3 2

Y 4 4 3

Destination

Neig

hb

or

Routing table of W


Handling errors

Transmission errors Signals are mutilated, not correctly converted to (intended) bits Local issue

Packets are missing Local or end-to-end issue

Overload problems Flow control: Fast sender overruns slow receiver Congestion control: Receiver would be fast enough, but sender

injects more packets into network than network is able to handle

Where and how to handle these errors?


Structuring communication systems

To handle complexity, partition into well defined subsystems

Called layers

Subsystem/layer offers a well defined functionality, a service

A promise on what will happen after certain interactions with the service access point take place

Services need distributed implementations Parts at sender, parts at receiver, and possibly parts in the network

These various implementation parts interact with each other via protocols

Rules on interaction, how to achieve the service’s promise


Typical examples of services

Datagram service Unit of data are messages Correct, but not necessarily complete or in order Connection-less Usually insecure/not dependable, not confirmed

Reliable byte stream Byte stream Correct, complete, in order, confirmed Sometimes, but not always secure/dependable Connection-oriented

Almost all possible combinations are conceivable!


Analogy: Nested layers as nested translations

Layers rely on services of lower layers

Intermediate representation of data changes

Vertical vs. horizontal communication

Vertical: always real

Horizontal: may be real or virtual

Horizontal communication



Horizontal (real!) communication

Verticalcomm.


Protocols and FSMs

Finite state machines describe & implement actual behavioral rules of a protocol

Have to communicate with their remote peer Cannot do so directly, have to use service of the underlying communication

layer Via service primitives, which can also provide arriving data to the protocol E.g., indications from lower layer are events to higher layer protocol engine

Layer n

Use service primitives

Layer n+1


Protocols and messages

When using lower-layer services to communicate with the remote peer, administrative data is usually included in those messages

Typical example Protocol receivers data from

higher layer Adds own administrative data Passes the extended message

down to the lower layer Receiver will receive original

message plus administrative data

Encapsulating Header or

trailer Layer n

Layer n+1 Layer n+1

Packet arriving at

Layer n+1’s SAP

Extended packet passed

to layer n

Delivered by layer

n


ISO/OSI 7-layer reference model (complete network)En

d-to

-en

dC

hain

ed

, local la

yers


TCP/IP protocol stack

Nothing statedby TCP/IP model


TCP/IP – Suite of protocols

Over time, a suite of protocols has evolved around the core TCP/IP protocols

So-called “hourglass model”: Thin waist of the protocol stack at IP, above the technological layers


ISO/OSI versus TCP/IP

ISO/OSI: Very useful model, non-existing protocols TCP/IP: Non-existing model, very useful protocols

Hence: Use a simplified ISO/OSI model, but treat the TCP/IP protocol stack in the context of this model

With suitable add-ons especially for the lower layers


Conclusion

These are the core functionalities a communication network has to provide

We shall investigate the main protocols of the abbreviated reference model in the remainder of this class

computer networks group universität paderborn computer networks chapter 1: a brief kms recap holger...

Documents