WAN Technologies & Topologies

Lecture 8
October 4, 2000
WAN Background
 WAN - Wide Area Network, spans multiple
cities/states.
 MAN - Metropolitan Area Networks exist in a
single city.
 LAN – Single building environment.

 Multiple MANs can make a WAN, multiple

LANs can make a MAN, etc.
WAN Example
MAN As A Building Block

 As I mentioned earlier, multiple MANs usually

make up the WAN.
 Each MAN has control of its own domain, but
it’s uplink to the other MANs is considered the
WAN (backbone).
 Using Verio Boston’s example…
The Interconnects

 MAN & WAN circuits must terminate in some

form of a packet switch.
 Using the UNet example from the other day,
the packet switches were the Cisco Catalyst
6509 switches.
 But keep in mind, the telephone company also
has some switches in the field that handle even
more traffic than what you’re getting!
Switch Interconnects – The “Cloud”
WAN Switch Functionality
 WAN switches use store and forward
technology.
 The store operation occurs when the packet
arrives: the I/O hardware copies the packet,
sticks it in memory, and signals the processor
to forward the packet.
 The forward operation is the act of removing
the packet from memory, and sends it to the
appropriate interface.
WAN Switch Functionality (cont.)

 Storing the packets also leads to a form of

queuing for each interface.
 If the destination interface is busy, the packet
is queued until the destination interface is idle,
then the forward occurs.
 The store and forward paradigm allows to
handle the maximum bandwidth of the WAN
connection, since all data is buffered!
Physical Addressing in the WAN
Environment

 A hierarchical scheme is used with WAN

addressing.
 The simplest form of this scheme: The first part
of the address holds the destination switch, the
second part holds the specific machine that the
packet is destined for on that switch.
 This is scheme is used in many WAN
environments.
WAN Addressing (cont.)
Next-Hop Forwarding

 In order for networking to occur, each device

must have some knowledge of the devices
which it is connected to.
 Next-hop forwarding is a scheme where
devices know their neighbors, but don’t know
the specifics of what is connected to each
neighbor.
The Airline Example
– Suppose a passenger is traveling from San
Francisco to Miami. Only one flight is listed, with
three legs: Dallas, Atlanta, Miami.
– From San Francisco, his next destination is Dallas.
– From Dallas, his next destination is Atlanta.
– From Atlanta, his last destination is Miami.
– But all along, the LAST destination was Miami, even
though the next hop was changing at each leg.
Next-Hop Diagram
Source Independence

 The next hop does not depend on the direction

that the packet came from.
 This is referred to source independence.
 Source independence is a fundamental
concept in data networks. It allows for low-
overhead, efficient networks.
Hierarchical Addressing & Routing

 Heirarchical addressing is almost routing…

 The act of forwarding a packet to the next
address is dubbed routing.
 Routing uses a table format to determine the
next hop of the communication, and since it
only needs to inspect the first part of the
address, it is efficient!
A Simplified Routing Table
Routing (cont.)

 The two part addressing scheme provides us

with the following:
– Switches along the path of the transmission need to
inspect the first part of the hierarchical address.
– The last switch in the transmission must inspect the
last part of the address.
– Really efficient network transport! Not much
overhead.
Routing in the WAN
 There are LOTS of routing algorithms that are
commonly used in the WAN environment.
– RIP: Router Information Protocol
– OSPF: Open Shortest Path First
– BGP: Border Gateway Protocol
– IGRP: Interior Gateway Routing Protocol
 I’m not going to focus on the different
algorithms, but please read up and understand
the differences between them!
Routing in the WAN (cont.)

 The best way to visualize routing is to imagine

how all of the networks are connected
together.
 Each node in the network is a packet switch.
 Each connection between switches is a link or
an edge.
A Network Diagram
Routing Table for the Example
Default Routes

 Default routes allow for the simplification of

routing tables.
 Since many packets would have the same
routes, a default route would reduce the
amount of work the router/switch would have to
do.
Simplified Routing Table
Default Routing (cont.)

 Only one default route per device is allowed.

 The default route has lower priority to other
entered routes.
 If a transmission does not find a valid route, it
will send the packet down the default route.
Determination of the Routing Table

 Two ways exist for route determination:

– Static routing
– Dynamic routing
 Why have different options?
Static Routing

 Static routing is the most straightforward of the

two schemes.
 Pros:
– Simple to visualize
– Low overhead on devices which perform routing
 Cons:
– Static, inflexible
Dynamic Routing

 Dynamic routing can be very confusing.

 Multiple types of dynamic routing exist:
– OSPF
– IGRP
– BGP
– RIP
 Main disadvantage of dynamic routing: difficult
to understand.
OSPF
 OSPF – Open Shortest Path First
 Uses a system of metrics to determine route
preference.
 The entire route preference is the sum of the
individual metrics of the links between the
computers.
 The routers send their routing tables out
periodically to their neighbors.
Common WAN Technologies

 ATM- Asynchronous Transfer Mode

 Frame Relay
 SONET
ATM
 Data is divided into small, fixed packets called cells.
 Each cell is 53 octets:
– 5 octets for header info
– 48 octets for data transmission
 ATM originally designed for simultaneous transmission
of data, voice, and video.
 Quality of Service (QOS) is adjustable with bandwidth
needs: higher for video, lower for data and video.
ATM (cont).

 Connections are usually 155Mbps OC-3.

 Uses the “cloud” concept.
Frame Relay

 Data is divided into small, variable sized cells,

up to 16,000 octets!
 Up to 1.544Mbps transmission rate.
 Uses “cloud” concept – shared bandwidth with
other connections.
OSPF Example