CCNA 1


ieee 802.3 csma/cd ethernet

802.5 token ring

media access control

48 bit MAC addresses

1st 6 hex digits = vendor

last 6 hex digits = organisationally unique identifier

eg: 00-D0-41-54-AF-EE

if x AND y = 1 then f =1

2048

1024

512

258

128

64

32

16

8

4

2

1



128

196

228

244

252

254

255


carrier sense multiple access/ collision detection: if no tx is detected by a node it will tx

star

ring

bus

mesh

hierarchical

24 bit ip address

11111111

11111111

11111111

11111111

broadcast

each host sends data: 1st come 1st serve, eg ethernet

token ring

host only sends when it' has been given a 'token', eg FDDI


LAN

WAN

MAN


(fibre distribution data interface)

max no. peer to peer hosts: 10

intranet: a private-use wan

extranet: same but some outside secure users allowed


bandwidth: the amount of data that can be sent through a network connection. a finite resource.

throughput: the actual amount of bandwidth being obtained, reduced by type of data being sent, no. of users, power, topology, device and computrer variables.

best transfer time = file-size / bandwidth

vertical cabling: backbone to backbone connection media

horizontal cabling: 'normal;' connection media


OSI model:

application



tcp model:



application

(open system interconnect)

presentation




session





transport


segments


transport

network

routers

packets


internet


datalink

switches, nics

frame


network

access

physical

hubs/ repeaters



encapulation: lower layers add informational headers ie 'wrap' the data already there

'10 base T' =10Mbps: baseband ie not modulated: twisted-pair


10 base 2

thin coax

185m

'thinnet'

STP = shielded twisted pair

10 base 5

thick coax

500m

'thicknet'

ScTp = screened twisted pair

10 base T

twisted pair

100m

very cheap

UTP = unshielded twisted pair

10 base FL

fibre

400m



100 base TX

4 pairs cat5

100m


802.3u: fast ethernet

100 base FX

2 optic fibres (tx,rx)

412m


straight thru for like to unlike

100 base T4

4 pairs cat3

100m


crossover for like to like

1000 base T

4 pairs cat5e

100m

full duplex

802.3a,b: gigabit ethernet over UTP

1000 base CX

used in closets

25m

'twinax'


1000 base LX

62.5µ

50µ

440m

550m

10km


802.3x: gigabit ethernet

(the 1st three above are 'legacy' ethernet)





5 4 3 rule: FIVE network segments connected using FOUR repeaters with only THREE segments having hosts.

round trip delay times for 10 base T: repeater 2µs: nic 1µs: cable 0.55µs /100m

ISDN

(integrated services digital network)

data 64kps, data 64kps, sig 16kps (user,user, terminal)

BRI: basic rate interface

WANS

use serial tx'ers. DTA's at users end, DCE at providers end.

(data terminal equipment)

(data communications equipment)



half duplex: one only node can tx

asynchronous: no timing data (but is there for compatibility

full duplex: more than one node can tx: no collisions

synchronous: timing data is used eg in 10 base x

frame

MAC layer in the datalink layer , max 1500bytes, uses mac addresses, all nodes look at the MAC address of incoming frames to see if it relates to them. If a frame is in error or not received the rx does NOT send an ack, and the tx knows to resend the frame, after 16 attempts to send, the MAC layer signals an error to the network layer


header: control information sent at the start of a

jam: a signal sent after a collision to stop any node tx'ing

broadcast domain: all the nodes that a broadcast frame will reach: stopped by routers

collision domain: all the nodes that will sense a collision, stopped by bridges, switches and router

a collision detected local collision: a collision detected by both tx and rx: both know to stop

backoff time: the random time, in multiples of slot time, that a tx waits before tx'ing again

remote collision: a collision on the far side eg of a repeater, it will have no FCS and be seen as a short frame

late collision: a collision which got detected as such after 64 bits of the data has been sent, it will be lost.

slot time: the max duration of a frame, to ensure collisions are detected, taking into account round trip delay time

collision:

(runt): simultaneous tx before slot time has elapsed


late: simultaneous tx after slot time has elapsed


long frame: tx exceeds frame size (jabber)


short frame: collision fragments <64 octets


alignment/ range error: insufficient/ excessive bits, no. of octets misreported


FCS error: corrupted frame


ghost/ jabber/ XT: long pre-ambles, very long tx's, noise, crosstalk

link pulse:

timing data sent to link different speed ethernet lans and hence allow auto-negotiation

packet starvation :

the tx which sends successfully is liable to carry on doing so, SNMP must be used to give other hosts connectivity again

consumption delay:

wasted bandwidth due to collisions

enforced idle time:

wasted bandwidth caused backoffs

frame forwarding:

a layer 1 hub WILL send: a layer 2 switch MAY send, a level 3 router will TRY NOT to send

mac address flooding:

a node 'floods' the broadcast domain to try and find it

spanning tree protocol:

a protocol used by switches to stop 'floods' from looping around

microsegmentation:

the use of switches to reduce the size of collision domains

layer 2 devices build lists of MAC addresses and maps them to the correct outlet or 'port'

a bridge is in effect a layer 2 repeater

switch modes:

store-and-forward: reads entire frame, error checks, sends. slow, reliable


cut-through: reads rx mac address, sends. fast, uses synchronous switching


fragment-free: reads 1st 64 bytes, ie sends if not a collision, uses synchronous switching

spanning tree protocol:

at the logical link layer, sends BPDU's (bridge protocol data units) giving peer values to switches and creating a somewhat 'hierarchical' network


blocking:

does BPDU's but does not forward frames


listening:

does BPDU's, builds a MAC address table, does not forward frames


forwarding:

does BPDU's, forwards frames


disabled:

just switched on , does no BPDU's or forward frames


initialise: block: listen: learn: forward or disabled

ARP:

sends requests for MAC addresses

RARP:

send requests for IP addresses


BOOTP:

uses UDP to give out client IP addresses

DHCP:

advanced protocol based on BOOTP


multicast:

the use of class D network address to communicate with many hosts without creating seperate sessions. CGMP and IGMP is used to prevent bandwidth reduction.

UDP:

connectionless, eg TFTP. uses application layer protocols to get the reliabilty

TCP:

a connection-orientated protocol ie it creates a 'virtual circuit' before data is sent

IGP:

interior gateway protocol, common admin control for an autonomous network: rip, ripV2, igrp, eigrp, ospf, isis

rip uses hop count. max 15 hops.

ripV2 advanced rip

igrip uses bandwidth, delay, reliabilty. load and remembers 256 routers



EGP:

exterior gateway protocol, same between other networks, bgp




BGP: border gatewayprotocol

distance vectoring protocols:

rip, igrp, eigrp

link state protocols:

ospf, isis



port numbers:

20, 21 FTP

23 telnet

25 SMTP

53 DNS

67, 68 BOOTP


69 TFTP

80 HTTP

>1024 'dynamic'