Well,
failed by 1 questions.
I guess I'll take again in a few months.
Questions were very poorly worded but what can you do.
Monday, March 4, 2013
Saturday, March 2, 2013
Network management
Alright last big chapter.
Network management is done at the last two stages.
Operate, then based on the data accumulated the network will be
optimized...
Statistics will be grabbed on
link utilization - to see if the link needs upgrading
CPU utilization - to see if the device will need upgrading
interface utilization - to see traffic patterns
Cisco Works can centralize configuration changes.
FCAPS is the acronym for this chapter.
Fault management
Configuration management
Accounting Management
Performance management
Security management
This is all done using.
NMS - Network management Systems. This is not a product but a framework for Network management.
In order to run this management the NMS will need.
Network management protocols and standards. - These will be SNMP, RMON,
The NMS will manage devices - these will be the "managed devices"
Inside each device that is being managed there will be a diplomat a "Management Agent".
Snmp Agent
RMON Agent
Netflow can be used instead of RMON.
Syslog can accumulate data from all the devices instead of storing the logs/data locally on each.
SNMP RFC1157
Runs over UDP.
The data in each device is stored in it.
The storage is arranged in a TRee format.
MIB.
The MIB can be queried for the data in the cell/branch by referencing its locations
either by name or number.
Name
Syntax - interger or string
Encoding
Normal Tree would give you basic data like the interface packets.
Cisco has some "private" MIB which will give you the small, med , larg packets.
SNMPv1
Request and Respond mainly.
Get request - let's see get CPU usage - (get response 60%)
Get Request - let's see get route table -(get response 10.0.0.0/24 next-hop 15.15.15.1)
Get Next Request - get response 20.0.0.0/24 next-hop 16.16.16.1)
when the agent responds to the requests he will send a Get response
okay so far we have been reading the fields
Set Request - will enable you to write to a field. set request mib3.3.3.4.5 contactname Saar
pretty useless, I can't find any real implementations of this.
Trap - this is a setting on the agent. When a certain item on the device happens it will try to send an
alert about it to the NMS.
For example on linkdown send to NMS.
SNMPv2
added getbulk that way you don't have to repeat the getnext requests
added inform request basically an improved trap with Confirmation.
SNMPv3
This one finally adds Security.
noAuthnoPriv no authentication at all and no privacy (which means no encryption)
authNoPriv ok, authentication is ok but no privacy. (no encryption)
AuthPriv authentication and Privacy.
Great!!!!
Authentication is MDAC
Encryption is DES, 3DES, AES
RMON
Network Nodes are needed for this.
Can't avoid it.
They are expensive. I think the leader in this is NetScout.
Looks at MAC Layer 2 data.
RMONv1 grabs data from the Layer 1-2
RMONv2 grabs it from the layer 3-7
Netflow is a higher level of the above.
It grabs.
Accounting is the data grabbing
Collectors will grab the data physically.
Analazyers will give you the reporting and GUI for this.
The netflows can be used for billing.
Network planning
Planning for user actions
or Application actions.
CDP
this is a Cisco protocol.
Cisco Discovery Protoocl.
It is helpful for troubleshooting
It runs on Layer 2 level.
syslog allows you to get information from multiple sources.
You can accumulate it all on the Syslog device and then use that to get data.
The levels go from the lowest
Which is the most critical.
0 Emergency
to 6 infomrational
7 debug
Friday, March 1, 2013
Voice and Video
PSTN is circuit switched. Which means that the circuit is built and used for the entire
connection. There is no switching done while the circuit is alive.
CO central office use SS7 in order to route and build the circuit.
The call can be build on Dialup , ISDN , or a TDM.
Each call is 64 kbps of bandwidth and is called a DS0
DS0=64 Kbps.
Old PBXs sit in the enterprise and will give you.
Extension dialing.
VoiceMail
transfers
conferencing
To connect to another site a company can set up a TIE line .
On the TIE line there are no charges to the enterprise.
However the TIE line itself costs money.
Alright, on net which is on the TIE lines.
Off NET to the PSTN.
This will be the same even in VOIP.
If you are using T1's you will call it On-Net
If you are having a call over the internet it will be off-net
PSTN requires you paying charges per each call.
While a Tie line has a fixed monthly cost.
T1 can carry only 24 calls.
24 * 64 = 1280 256 = 1536 Kbps
Now in most books they say T1 is 1.544 Mbps .
So where are the missing 8 bps.
Apparently those are used by the telco for synchronization.
So 23 B channels + 1 D channel = 24 Channel then you need to add 8 Kbps for synchronization.
In the case of Telco. This is ALL used for calls and cannot be used for Data.
Ok.
CO - central office.
This is a map of all the CO (central office) in the USA.
Notice how the west coast has less per mile.
So I drew this up. Since the CCDA one in the book looks pretty useless.
So Tie line is what I buy so I can connect two offices and not pay toll.
Tandem Trunk is what the PSTN provider uses to connect local CO (local exchanges is the correcter term)
Tandem trunks go to a Tandem Switch (class 4)
They will connect to a Class 3 switch which will connect to another Class 3.
Technically if you want to dial abroad then you need to reach a Class 1 switch.
Anyway from Switch 3 to Switch 3 it is and INTER TOLL trunk.
Co to PBX and PBX to Co is just the connection to the CO from the office.
As a note. When you dial in NY. you only need to dial 7 digits since it uses the TANDEM trunks.
When you dial to boston you need the FULL number. it goes on the TOll trunk.
Okay
FXS
Foreign Exchange Service.
We are Exchanging - ie TALKING
In VOIP you will use the same ports to connect OLD equipment to your VOIP network.
Like the ATA from Cisco gives you two FXS ports for the old devices.
So FXS ports
provide Dial tone
Power
Ring Voltage.
Now
FXO is the port that
Ah fuck it.
Just try this.
FXS point to the STATION.
FXO points to the central Office.
So on the phone you have an FXO port
you plug the cable to the FXS port on the PBX
The PBX has an FXO port
that you will plug a cable from the FXO port
to the FXS jack which is the POTS circuit to the Central OFFICE.
E&M ear and mouth - Earth and Magnet.
This is basically a PORT on a PBX. you run an Analog cable which will run to another PORT on PBX2
This allows you to send a signal. This is a TIE trunk for analog.
This has been replaced by BRI PRI digital.
Since we use T1/E1
T1 has 24 channels. It can work either.
CAS Channel associated Signalling - The signalling here is in each channel
In each channel a bit will be robbed for signalling. So 24 channels.
CCS - common channel signalling - This uses one channel for signalling so 23B+D
ISDN uses this and so does SS7
Signalling the state of the phone.
Supervisory signalling tell if it is on hook or off hook
Addressing sends the digits.
Informational sends you the BUSY
Loop start - residential CO to Phone. When you lift the handset the circuit is closed.
Ground Start - CO to phone signals to the switch that it is about to take the line.
helps prevent glaring which is when both take the line at the same time.
E&M - PBX to PBX Two wire - four wire adds more signalling
CAS T1 occurs in band
CSS T1 sets up a separate channel for the signalling
QSig Q.931 used for ISDN between PBX to PBX and Hybrid to CUCM
SS7 inter PSTN switches signalling used by the PhoneProvider.
Loop Start
the CO has the Power 48 DC.
That is why a phone does not need electricity.
When you lift the handset the circuit is closed (off hook) and the power flows all the way to
the phone and back to the CO which sends a dial tone.
Ground Start.
Uses TIP and RING
The PBX has a TIP detector
When the CO grounds the TIP
the PBX detects this and will ground the RING.
Now the CO power 48DC can flow and the arrival of the 48 DC will signal to the CO to send Dial tone.
If the PBX wants to ring.
It will ground the RING which will be detected by the CO.
The CO will ground the TIP
Now the CO power 48 can flow and when it reaches the CO it send Dial tone
E&M
type I and type II are in the USA
type III is everywhere.
Type V is outside the USA.
Immediate start wait 200ms and send
Wink - wait for on hook from other side and then sends.
Delay start - lift wait for 200ms check if on hook. if still off-hook then wait till on-hook.
on hook means
Off- hook means
SS7 signalling
So
Best I can do STP would be internal.
SCP for control - controls 800-900 and credit cards
Sending the Dialing Numbers
Pulse - rotary - sends 5 signals 1 1 1 1 1 = 5
DTMF - sends Two tones at the same time. X+Y = 5
NANP
North American - numbering plan
country-area-office-line
nxx-nxx-xxx-xxxx
Centrex - the CO has the PBX
VoiceMail -
Database or CDR keeps data on the calls. You can get reports.
IVR - interactive for accounting press 5 for Jamil press 2
ACD - used for call centers - how long the calls + statistics.
Grade of service.
P.02 GOS = means 2% of calls won't make it
P.01 GOS - means 1% won't make it and will be blocked.
Earlang B.
You have 24 channels
if one is taken for a whole hour = 1 Earlang.
You measure it at the busy hours.
30 calls * 10 minute per call = 300 minutes 300/60 = 5 Earlangs.
So you run Earlang with the GOS to calculate the number of lines you need.
Earlang B extended
Same thing but takes into account some people will retry to call.
Earlang C
Same thing but instead of blocking calls will put them on HOLD.
Used for C callcenter.
Call second = 1 second of calls.
Centrum = cent 100 seconds
1 Centrum * 36 = 1 Earlng
Busy Hour also called Peak Hour
Busy Hour traffic.
average call duration (20 minutes) * number of calls (6) = 120 minutes/60 = 2 Earlangs.
If you have the length of the calls in seconds.
Then you need to divide it in seconds. So instead of 60 minutes you use 3600 seconds.
So an example.
each call is 300 seconds * 400 calls = 120,000 seconds / 3600 seconds = 33.333 earlangs.
Eventually you will have blocked calls. They are the GOS you plan for.
CDR
call detail records done by the PBX
ACD distributes calls to agents.
Cisco Unified Network
Network is at the base.
Call routing done by the CUCM / SRST dial plans and PST gateway
Call Control on off to the LDAP
Applications and services like chat or voicemail or contact center
Operations and Service Quality
VoFR - over frame relay
VoATM - over ATM
VOIP is now the leader.
Voice over IP.
PSTN is not flexible
Data networks are
You can also save money by adding it all on the DATA network.
SRST - backup goes to PSTN . Overflow of traffic goes to PSTN.
Cisco IPT IP telephony.
Network is the base.
Gateways convert from VOIP to analog
CUCM is the brains of the dial plan and routers VOIP
Single Site - one CUCM at HQ
Multi site CENTRALIZED - CUCM at HQ and SRST at branches
Multi site Distributed - CUCM at each location .
In a multi-site distributed.
Each Cluster up to 30000 users.
To connect clusters use Inter-cluster trunk
GateKeepers will keep them in sync and enforce CAC
CAC is call access control. Which prevents too many calls on the line. So the call quality is kept up.
Call manager express is one appliance that does the mailbox, pstn and voip
Video
Telepresence which is the Cisco Video conferencing is many to many and takes 4-12 Mbps HD
Desktop Video to Video - many to many less demand
cameras to hq - many to few many cameras send data to few(hq)
Streaming Video Few to Many - the VOD to many users.
Access for users
Transport for sending the data
Bridging for converting it.
session provide signalling.
Storage store content
Codecs
G711 8000 samples per second * 8bits per sample 64000 = 64 Kbps = DS0
Analog to Digital
Filter the range you want to record. anything above 4000hz will be dropped
sample the speaking 8000 times per second
Digitize it into 0 1 01 1 01010 1 also called PCM pulse to Code modulation
G711 U for USA 64 Kbps
G711 a for international 64Kbps
G729 8 kbps
G728 16 Kbps
G726 16-40 Kbps
G723.1 6.3 5.3 Kbps
711 is the best then 729 728 726 723.
If you have the bandwidth keep it at 711
if you want to compress and not lose quality 729
Call control
Q931 for ISDN
H225 for the rest.
This is done over the TCP
UDP
G7xx is the Audio over UDP
H.26x H for Hvideo
or RTP real time Protocol does them both.
RTCP is control of the Video/audio
RAS is control
SCCP Cisco proprietary VOIP call cotrol..... IT only sets up the Control
RTP for voip streaming. this is the actual data call stream.
MGCP media Gateway Control Protocol
The HQ Gateway controls everything.
SIP - voip networks for non-cisco proprietary.
IP = 20 bytes
UDP = 8 bytes
RTP = 12 bytes
cRTP compresses the 40 Bytes to 2-4 Bytes
Hop by Hop for small 768 kbps links
MGCP allows the CUCM to control Gateways that go to PSTN
the CUCM is the Call agent endpoints are the phones.
H.323
Terminals are the clients
MCU mixes streams
Gateway converts to PSTn
Gatekeeper - Dial plan + CAC used for multisite distributed CUCM
Gatekeeper works like a OSPF DR and holds the Dial plan.
SIP proxy manages the SIP clients
VAD supresses silence
Propagation delay is the travel time.
Processing delay is the time to convert it to digital
Serailization is how long to put it on the interface. -
Queuing delay is waiting ebcause of other packets. - LFI and QoS helps
Jitter is the change in the delays - use dejitter buffers
Echo delay of 15ms and above must be cancelled.
Classify MATCH
Mark it with the color THEN
Congestion avoidance by using WRED or DWRED to drop tails.
Traffic Policing
Traffic Shaping by buffering and releasing slowly.
AUTO-QoS
Marks
does 802.1Q
LLQ
CBWFQ for control traffic
P>S.
VOICE sucks.
Wait till you have to use the Cisco tools to sell someone a Callmanager.
IPv4
IPv4
Version
0100 = 4 so IPv4
0110 = 6 so IPv6
IHL internet header length.
How long is it in Bytes (IPv4 changes size) (IPv6 is fixed)
TOS Type of service which is your QoS DSCP coloring marking.
Total Length of the packet including the data. Useful for determining if you need to fragment the packet
Identification - identifies the Fragments.
Flags
0 Fragment
1 do not fragment
Fragment offset this is fragment 1 of 40
TTL time to live each hop cuts 1
protocol 8 bits used by IANA
1 ICMP (ping)
2 IGMP (multicast)
6 TCP
17 UDP
50 ESP
51 AH
88 EIGRp
89 OSPF
103 PIM (multicast)
112 VRRP
Header Checksum - used to see if the packet is still ok after transport. Changes every header change.
Source address 32 bits
Destination address 32 bits
IP options not in use. Used for security , route record and similar.
Padding so the packet ends on a 32 bit boundary.
TOS
used for the QoS.
Voip is 101
nothing is 000
PIFFCIN priority imediate flash flash Critical in network
TOS itself is 4 bits.
It can be used to select a route based on.
Money
reliable
throughput
delay
With DSCP they dropped the TOS which nobody used.
The way it works now is Class 4 will have a higher priority.
and if there is congestion the high Drop will be dropped first.
Precedence AF 1 AF 2 AF 3 AF 4
Low drop precedence 001010 010010 011010 100010
Medium drop precedence 001100 010100 011100 100100
High drop precedence 001110 010110 011110 100110
So a 1 at the beginning is better. 1xx
A 010 will not be dropped.
MTU ethernet 1518
LAN jumbo frames
TCp will retransmit
UDP wont'
Class A 0xxxxxxx so 0 to 127
Class B 1xxxxxxx so 128 to 191
Class C 11xxxxx so 192 223
class D 111xxxxx so 224 239 multicast
Class E 1111xxxx so 240 to 255 experimental.
Unicast
Broadcast
Multicast
Private are not routed 10/8 172.16/12 192.168/16
1 class A 16 Class B 256 Class C
You can subnet the addresses above.
Static nat is ONE to ONE private to Public.
Dynamic NAT overloading is PAT port Address translation.
Dynamic Overloading is an internal pool to an external one.
Inside Local is the IP of my PC.
Inside Global is the Public IP I get on the web
Outside Global is the IP of a device on the WWW.
Outside Local is his IP when he is in my STUB/LAN
BOOTP
get IP and gateway using UDP replaced
DHCP
Manual is to map a MAC to an IP address.
Automatic does not expire
Dynamic is from a Pool and expires.
DHCPDiscover.
Router can relay this
DHCP Offer
DHCP request
DHCP acknowledge
DHCPNAK not acknowledge I am out of addresses.
DHCP should be in the server farm / datacenter
Internal DNS campus
Edge External
remote datacenter BOTH
ARP
IPV6
128 Bits instead of 32 bits per address.
Each IP is globally unique
Header is fixed at 40 Bytes
Header will reference options so it is a fixed size.
Addresses can autoconfigure if required.
IPSEC is built in
MTU discovery
Multiple IPv6 addresses
No broadcast replaced by multicast
Version 0110 IPV6 = 6
Traffic class 8 bits = TOS
Flow 20 bits for ordering the flow.
Next Header to add more
Hop limit = TTL
source
destination
6
17 udp
50 esp
51 ah
88 eigrp
89 opsf
ipv4 compatible 000000000x.x.x.x
FF multicast
FE link local
FC private addressing
Global is routable
64 bits netowrk 64 bits host (made up from the MAC 48 bits)
To convert a MAC 48 to 64 you add two FF FF in the middle.
FE is link local can be auto configured
FC is private addressing Unique
Globally aggregetable = aggregate of the IPV6
Anycast is to the nearest.
FF:01 1 all nodes
FF: 01 2 all routers
FF:02 5 OSPF
FF:02 6 OSPF designated
FF:02 9 RIPnG
FF:02 A EIGRP
FF:02 C DHCP
ICMPv6 discovers MTu
IPv6 ND neighbor discovery
IPv6 DNS AAAAAAAAAAAA
Use the same DNS server.
Stateless link local
Stateless global
Stateful DHCP
Global
Talk to router and gets the prefix
Prefix + MAC = address
EIGRP for IPv6
RIPnG
OSPFv3
BGP4
ISIS for IPv6
Dual Stack is both IPv4 and IPv6
Tunneling IPv6 into an IPv4 tunnel
Translate IPv6 to IPv4
Dual Stack - if DNS sends AAAAA it uses IPv6
Automatic Tunnel
IPv4 compatible
6 to 4 the destination has an IPV4 in it which is used as the tunnel envelope
6 over 4 Multicast over Multicast
ISATAP - Greek, Chinese
Daul stack can use PAT or NAT-PT
Ciscio 6PE over MPLS
Service Block service the translations.
Version
0100 = 4 so IPv4
0110 = 6 so IPv6
IHL internet header length.
How long is it in Bytes (IPv4 changes size) (IPv6 is fixed)
TOS Type of service which is your QoS DSCP coloring marking.
Total Length of the packet including the data. Useful for determining if you need to fragment the packet
Identification - identifies the Fragments.
Flags
0 Fragment
1 do not fragment
Fragment offset this is fragment 1 of 40
TTL time to live each hop cuts 1
protocol 8 bits used by IANA
1 ICMP (ping)
2 IGMP (multicast)
6 TCP
17 UDP
50 ESP
51 AH
88 EIGRp
89 OSPF
103 PIM (multicast)
112 VRRP
Header Checksum - used to see if the packet is still ok after transport. Changes every header change.
Source address 32 bits
Destination address 32 bits
IP options not in use. Used for security , route record and similar.
Padding so the packet ends on a 32 bit boundary.
TOS
used for the QoS.
Voip is 101
nothing is 000
PIFFCIN priority imediate flash flash Critical in network
TOS itself is 4 bits.
It can be used to select a route based on.
Money
reliable
throughput
delay
Class 1 (lowest) | Class 2 | Class 3 | Class 4 (highest) | |
---|---|---|---|---|
Low Drop | AF11 (DSCP 10) | AF21 (DSCP 18) | AF31 (DSCP 26) | AF41 (DSCP 34) |
Med Drop | AF12 (DSCP 12) | AF22 (DSCP 20) | AF32 (DSCP 28) | AF42 (DSCP 36) |
High Drop | AF13 (DSCP 14) | AF23 (DSCP 22) | AF33 (DSCP 30) | AF43 (DSCP 38) |
With DSCP they dropped the TOS which nobody used.
The way it works now is Class 4 will have a higher priority.
and if there is congestion the high Drop will be dropped first.
Precedence AF 1 AF 2 AF 3 AF 4
Low drop precedence 001010 010010 011010 100010
Medium drop precedence 001100 010100 011100 100100
High drop precedence 001110 010110 011110 100110
So a 1 at the beginning is better. 1xx
A 010 will not be dropped.
MTU ethernet 1518
LAN jumbo frames
TCp will retransmit
UDP wont'
Class A 0xxxxxxx so 0 to 127
Class B 1xxxxxxx so 128 to 191
Class C 11xxxxx so 192 223
class D 111xxxxx so 224 239 multicast
Class E 1111xxxx so 240 to 255 experimental.
Unicast
Broadcast
Multicast
Private are not routed 10/8 172.16/12 192.168/16
1 class A 16 Class B 256 Class C
You can subnet the addresses above.
Static nat is ONE to ONE private to Public.
Dynamic NAT overloading is PAT port Address translation.
Dynamic Overloading is an internal pool to an external one.
Inside Local is the IP of my PC.
Inside Global is the Public IP I get on the web
Outside Global is the IP of a device on the WWW.
Outside Local is his IP when he is in my STUB/LAN
BOOTP
get IP and gateway using UDP replaced
DHCP
Manual is to map a MAC to an IP address.
Automatic does not expire
Dynamic is from a Pool and expires.
DHCPDiscover.
Router can relay this
DHCP Offer
DHCP request
DHCP acknowledge
DHCPNAK not acknowledge I am out of addresses.
DHCP should be in the server farm / datacenter
Internal DNS campus
Edge External
remote datacenter BOTH
ARP
IPV6
128 Bits instead of 32 bits per address.
Each IP is globally unique
Header is fixed at 40 Bytes
Header will reference options so it is a fixed size.
Addresses can autoconfigure if required.
IPSEC is built in
MTU discovery
Multiple IPv6 addresses
No broadcast replaced by multicast
Version 0110 IPV6 = 6
Traffic class 8 bits = TOS
Flow 20 bits for ordering the flow.
Next Header to add more
Hop limit = TTL
source
destination
6
17 udp
50 esp
51 ah
88 eigrp
89 opsf
ipv4 compatible 000000000x.x.x.x
FF multicast
FE link local
FC private addressing
Global is routable
64 bits netowrk 64 bits host (made up from the MAC 48 bits)
To convert a MAC 48 to 64 you add two FF FF in the middle.
FE is link local can be auto configured
FC is private addressing Unique
Globally aggregetable = aggregate of the IPV6
Anycast is to the nearest.
FF:01 1 all nodes
FF: 01 2 all routers
FF:02 5 OSPF
FF:02 6 OSPF designated
FF:02 9 RIPnG
FF:02 A EIGRP
FF:02 C DHCP
ICMPv6 discovers MTu
IPv6 ND neighbor discovery
IPv6 DNS AAAAAAAAAAAA
Use the same DNS server.
Stateless link local
Stateless global
Stateful DHCP
Global
Talk to router and gets the prefix
Prefix + MAC = address
EIGRP for IPv6
RIPnG
OSPFv3
BGP4
ISIS for IPv6
Dual Stack is both IPv4 and IPv6
Tunneling IPv6 into an IPv4 tunnel
Translate IPv6 to IPv4
Dual Stack - if DNS sends AAAAA it uses IPv6
Automatic Tunnel
IPv4 compatible
6 to 4 the destination has an IPV4 in it which is used as the tunnel envelope
6 over 4 Multicast over Multicast
ISATAP - Greek, Chinese
Daul stack can use PAT or NAT-PT
Ciscio 6PE over MPLS
Service Block service the translations.
Subscribe to:
Posts (Atom)