Transcript Slides for Week 5

Service Providers &
Data Link & Physical layers
Week 4 Lecture 2
To-day we will look at the
WAN and who provides it.
Telstra
AT&T
NTT
Optus
IP Links are router to router
Router
Switch
Some background
• The National and International telephone networks were
built for voice and circuit switching
• Connections to customers were analogue although interexchange links were being converted to digital
• Demand for data communications started in the seventies
• Data traffic now exceeds voice
• Privatisation of old monopolies started with the AT&T
break-up in 1986
Background continued
• Telco’s started to bring WAN to large
buildings – Fibre
• Telco’s expanded capacity dramatically in
the late 90s – 5 times the fibre X 100 time
efficiency = 500 times the capacity
• But usage is only doubling each year
• Hence the Telecom shakeout we are seeing
The larger Telcos
1. AT&T
US
2. Verizon
US
3. NTT
Japan
4. SBC
US
5. Worldcom (UUNet)US
6. Deutsche Telekom Germany
7. France Telecom
France
8. BT
UK
$66
$65
$58
$51
$39
$39
$32
$30
19. Telstra
$12
Australia
Source : Year 2000 company reports in US$bn
International Carriers
• Each of these have International “Hubs” in the
countries in which they operate
• They own or lease a share of the intercontinental
cables
• Many of them “own” the last mile and much of the
national networks
• They either provide ISP services direct to users
(Telsta’s Bigpond or UUNet’s Ozemail), or
• To smaller ISP that service geographical areas
The major carriers share cables
such as:
• PacRim East & PacRim West
– Telstra has a share
– 1 gbps transmission rate
• Europe to Asia link
• Southern Cross Cable between Australasia
and the US
• 600 communication satellites
Some characteristics of the
Southern Cross Cable
• Owned by:
• Telecom NZ (50%)
• Cable & Wireless Optus (40%)
• WorldCom (UUNet) (10%)
• Went live in November 2000
• Currently 240gbps and 480gbps eventually – that
is the data on a 3km stack of paper or in 8 full
length films, every second
• Reduced TransPacific costs to 1/10 and
transmission time from 300ms to 70ms
Southern Cross Cable cont.
• 30,500 km of cable in 2 separate cables
• Optical repeaters every 40-70km
• Designed to provide 99.999% reliability – that is
50 minutes downtime in 10 years, but it went
down early in 2001 as mods were being done to
one link and the other was cut by a ship
• Managed out of Auckland
• Connection points in Alexandria & Brookvale
Southern Cross Cable cont.
• The cable consists of:
•
•
•
•
•
•
•
Six strands of fibre
Set in jelly & hydrogen
Within a steel tube
Protected by high strength steel wire
Surrounded by seam welded copper
Insulated by high density polyethelene
To form a cable 18mm thick
Southern Cross cable
We tend to think the public
Internet and the Web are
dominant.
But most large organisations
have their own WAN.
More than 30 companies offer Frame
Relay or ATM services in Australia
• AAPT has Optic fibre in most capital cities with
fibre direct to 250 buildings. They have 50
FR/ATM POPs
• Cable & Wireless Optus have 38 FR/ATM POPs
• CITEC a business unit of the Queensland State
Gov.
• Telstra has 300 FR/ATM POPs and 2,200
customers
Figures as at September 2000, Telsyte report
An Organisation’s network can be:
•
•
•
•
•
Leased channels
VPN Virtual Private Network
VPN on Public network
Public Network
Combination of some or all or these
Data link & Physical Layers
• Function is to move frame over individual link
• Services can include
–
–
–
–
–
Point-to Point or Multicast
Reliable service
Flow control
Error detection & correction
Half or Full duplex
• Data link and Physical often implemented in the
adaptor or NIC
Wide variety of Protocols
• Many changes taking place
• Level at which services are sold thus as a
DIS architect you may need to negotiate for
these services
• Different LAN and WAN Protocols
• Earlier protocol families are being used at
this level
The main WAN protocols are
•
•
•
•
•
•
•
PPP – Point-to-Point Protocol
HDLC – High level Data Link Control
FDDI – Fibre Distributed Data Interface
ATM – Asynchronous Transfer Mode
X.25 & Frame Relay
SONET – Synchronous Optical Network
ISDN – Integrated Services Data Network
Point-to-Point Protocol
•
•
•
•
Typical protocol for link from home to ISP
Used over any medium – digital or analogue
Encapsulates datagram in a PPP frame
Has a link establishment & termination
procedure
• Detects errors but the protocol does not
require it to fix them
Asynchronous Transfer Mode
• Developed in 1980s by ATM Forum and ITU
• Can provide a full solution for a broadband,
packet switched virtual circuit network
• Widely used in telephone and Internet backbones
• With ascendance of TCP/IP likely to remain at
Data link level
• Carries IP using special interface called AAL5
ATM Continued
•
•
•
•
•
•
•
Investment has resulted in very high performance switches
Fixed length 53 byte packet carrying 48 bytes of data
Fixed length allows fast hardware switching
Bit level error detection & correction
Some link level congestion control
Different levels of Quality of Service (QoS)
At PMD (Physical Medium Dependent) Layer it operates
over copper, microwave & fibre
ATM over SONET
• In multiples of 51.84mbps
• Examples are:
–
–
–
–
OC-1
OC-3
OC-12
OC-192
51.84mbps
155.52mbps
622.08mbps
10gbps
• But these are multiplexed
– 160 channels of OC-192 gives 1.6tbps
– 82 channels of OC-768 gives 3.28tbps (under test)
An ATM network can look like one logical link
to IP with the AAL layer only at the edges of the
ATM network
IP
IP
AAL
AAL
ATM
ATM
ATM
ATM
Physical
Physical
Physical
Physical
ATM Switch
ATM Switch
X.25
• X.25 introduced in Europe in the early
1980s as the first packet switched network
• Because of high error rates on copper links
it provided for error recovery on a link-bylink basis
• Still in use in Europe
Frame Relay
• Introduced in the early 1990s as a second
generation X.25 network
• Designed for lower bit error rates & has no error
recovery & no flow control
• Lower overhead and thus higher transmission rates
and low latency
• Virtual circuit based, packet switching service
provided by Telcos
• Sold to corporate users as VPNs
Frame Relay (Cont.)
• TCP/IP and other protocols can be carried over Frame
Relay
• Variable length packets mean no hardware switching thus
slower than ATM
• Services up to 20mbps
• Sometimes numbers of FR circuits aggregated onto ATM
circuits and then back to FR
• Uses statistical multiplexing to provide:
– Permanent Virtual Circuits, or
– Switched Virtual Circuits
• Like ATM, Frame Relay networks can look like single
links to TCP/IP
CIR (Committed Information Rate)
•
•
•
•
VPN purchaser contracts for say 256kbps
Supplier commits to supply that rate
User can burst above that rate if unused capacity available
If capacity not available packets over the CIR may be
dropped
• This provides the guaranteed service level that TCP/IP is
unable to provide
• Users port access devices typically can deliver higher rates
than the CIR
SONET (Synchronous Optical Network)
• Standard interface for transporting digital
signals over optic fibre
• Allows multiplexing of circuits from
different vendors over the same fibre
• High speed rates as discussed before
• Circuit switched
• Usually only part of a carrier’s network
Rationalising the backbone links
IP
IP
IP
ATM
SONET
IP
ATM
SONET
WDM
WDM
WDM
WDM
WDM is Wave Division Multiplexing
Voice over IP will replace PBX and thus voice
will be carried over data networks
Evolution of private networks
• Based on leased lines/circuits
• ATM or Frame relay using permanent
virtual circuits
• Internet based VPNs or IP based VPNs
VPN (Virtual Private Network)
• Replacing private networks based on leased
circuits.
• A VPN is
– A WAN
– Providing what appears to be dedicated lines to
the organisation
– But are actually network facilities shared by
multiple customers
VPN (Virtual Private Network)
• Frame Relay service is the leading VPN service
• Now can be implemented as secure “tunnels” over
the Public Internet at the IP layer, between edge
routers or firewalls
• IP based VPNs are
– Encapsulated data paths
– Using one of a number of strategies e.g. IPSec (Internet
Protocol Security) to provide
• content encryption, and
• Packet authentication (prevent spoofing)
IP based VPN
• Because IP is layer 3, it gains the benefit of the
high up time service from ATM
• It is becoming popular because
–
–
–
–
Low cost
Simple to establish
Widespread connectivity
Access for mobile users
• ATM could allow service providers to offer the
SLAs required for corporate use.
IP based VPN (Cont.)
• Mobile users can dial into an ISP POP for a local
call and tunnel through to the corporate network
for e-Mail, Web and our corporate DIS
• IP VPNs can be used to extend Frame Relay VPN
for remote offices
• They can be used as Extranets – Secure
environments for B2B e-Commerce
• Extranet segregated from corporate network by a
firewall
Routing algorithms
The main function of IP at the Network layer
Routing algorithms
• The network layer has to determine the route the
message is to take
• In a virtual circuit all packets for the connection
will follow the same path
• In a datagram service like IP, packets may take
different routes
• In both situations the routing algorithm within the
Network layer will determine the routes
IP address = Network + Host
What does it do?
• It finds the best path from source to destination
• A good path is usually defined as least cost
• Where cost takes into account
– The number of hops
– The length of the hops and
– The congestion on specific hops
• But may also be affected by policy – e.g. do not
use routers belonging to Optus
Routing algorithms can be:
• Global – assumes all information about all
possible routes is known and the source to
destination can be calculated
• Decentralised – where routers exchange
information with their nearest neighbours
and slowly build up least cost information
They can also be:
• Static – where cost data only changes with
human intervention
• Dynamic – where the algorithm responds to
changes in topology and load on various
links
• The IP uses a dynamic approach
But the Internet has many routers
and hosts!
• It is broken down into AS (Autonomous Systems) where
all of the routers are under the control of a single
organisation – like Telstra or UUNet
• Routers in an AS use the same algorithm
• Protocols such as RIP are used within the AS and
• A special protocol called BGP4 is used when passing
packets between AS (in gateway routers)
• We also know that a service provider may push the IP
routers to the edge of its ATM network, which then uses its
hardware addresses for quick passing of frames through its
network
Autonomous regions
AS2
AS1
RIP
BGP4
AS4
AS3
IP Links are router to router
Router
Switch
These algorithms are complex
• Of great interest to mathematicians
• We do not need to know
• See Kuross page 301 for details if it excites
you
Intranet
• Uses Internet & Web technology: TCP/IP,
Browsers & Servers, HTML and HTTP
• To provide access to an organisation’s
information by staff in that organisation
• May use a private network or an IP VPN
Extranet
• Similar to an Internet, but access is given to
authorised users from specific organisations
• Typically used in B2B commerce
• Usually the organisation’s private Intranet
protected by a firewall
• Uses passwords and Digital certificates to
control access.