Mr. Mark Welton

thoughtlessskytopNetworking and Communications

Oct 29, 2013 (3 years and 5 months ago)

58 views

Mr. Mark
Welton


WAN
transportation method that formats data
into frames and sent over a network
controlled
by a service
provider


Frame
Relay is often represented in network
diagrams with a cloud depiction, representing
an unknown
environment


Frame
Relay uses VC (virtual circuits) through
the cloud to allow delivery to
endpoints


The
endpoints appear as directly connected
circuits


Two types of VCs


Permanent
(PVC
)


Circuit is always up


Path is “hard coded” through the provider’s system


Switched (SVC)


Are on
-
demand circuits


Path is created through the provider’s system when
used


Simple Frame Relay Network


Actual equipment involved in a frame relay
network


Each VC is given a Layer
-
2 address called a
Data Link Control Identifier (DLCI)


DLCIs
are only visible to the customer and the
service
provider


Other
customers of the service provider do
not see the DLCIs or other customer data
even though the Frame Relay network is
shared


The primary benefits of Frame Relay are cost
and flexibility


Dedicated
point
-
to
-
point
circuits (like T1s)
are priced based on distance between the
locations


Frame
Relay is
priced
based
on
the
components of the Frame Relay circuit


In
order to provision a Frame Relay circuit
four items are
needed


These four items will impact the cost of the
curcuit


Location for the circuit


Port
speed


is the physical size of the
circuit, such as a T1 or T3


CIR


Committed Information Rate


This is the guaranteed bandwidth
allocated by the service provider in bps


Burst Rate


is the amount of additional
bandwidth available on the
port




Burst Rate


is the amount of additional
bandwidth available on the
port


Typically this is ordered at 2 times the CIR or the
full port speed


If the CIR is exceeded but not the burst rate the
frames are marked as Discard Eligible (DE)


This means that if the frame relay switch
becomes congested they will be dropped


If the frames exceed the burst rate they are
automatically dropped






Frame Relay allows for multiple links to
multiple locations to terminate on the same
physical circuit and router interface


common designs for frame relay


Hub and spoke


Partial mesh


Fully mesh


All circuits are terminated to a central
location (usual the data center)


Branch sites can not communicate with each
other without going through the central
router


Design is commonly used when branches to
not need to communication directly


Similar to spoke and hub but some branch
site will also have circuits to each other


If two or more branch sites need to
communicate to each other often a frame
relay circuit is added directly between them


If an addition router was added to the right
design and only had a circuit to router A I
would consider it a partial mesh not spoke
and hub


In a fully meshed design frame relay reduces
the number of physical connections needed
vs. connections like T1s


To determine the number of links needed for
a fully meshed network the formula N(N
-
1)/2
is used where N is the number of nodes
(routers) in the network


How many links are needs for a network with
three router for a fully meshed design?


How many links are needs for a network with
three router for a fully meshed design?


3(3
-
1)/2 = 3


How many links are needs for a network with
six router for a fully meshed design? 6(6
-
1)/2
= 15


Using T1s you would need 15 T1


In Frame Relay you would need 6 Frame Relay
circuits with 15 DLCIs


Since
multiple locations can terminate on the
same physical circuit, oversubscription is
possible


Oversubscription
occurs when the amount of
bandwidth provisioned in all the CIRs exceeds
the port speed of the
circuit


CIR
cannot be guaranteed if it exceeds the
amount of bandwidth physically available


LMI


(Local
Management
Interface)
provides
communication between
the


Data
Terminal Equipment (DTE)


or customer
equipment like routers, and
the


Data
Communication Equipment (DCE)


or
service provider
equipment


LMI
provides an exchange of status messages
regarding the VCs


Three
forms of LMI are available on Cisco
equipment


Cisco


Ansi


Q933a


The
DCE device determines the type of LMI
used


Three
statuses of
PVCs


Active


normal good status for
PVC


Inactive


indicates the service provider has
configured a PVC, but the customer equipment is
not configured for that
DLCI


Deleted


indicates the customer equipment is
configured for a DLCI, but the PVC does not exist
from the service provider


Frame Relay networks will detect congestion
and mark frames with Forward Explicit
Congestion Notification
(FECNs)
and
Backward Explicit
Congestion

Notifications

(
BECNs)


They are
sent to the DTE
equipment


DTE equipment can adjust the flow of traffic
to reduce congestion on the
network


The DCE equipment does not perform flow
control for
customers


The FECN bits are marked when congestion
occurs


W
hen
FECN bits are seen in the frames, the
BECN bits get marked


Now traffic leaving and coming from a Frame
Relay switch is notified of the congestion


MPLS


Multi
-
Protocol Label Switching
networks are another common WAN
network


Like Frame Relay network diagrams represent
them as a cloud


packets in an MPLS network are prefixed with
an MPLS header (called
a
label
stack
)


The header contains one or more labels,
a
traffic
-
class field (used for quality of service
[
QoS
]), a bottom
-
of
-
stack flag, and


an 8
-
bit time
-
to
-
live (TTL) field



The label stack is the only thing examined by
the MPLS switches


no
traditional routing
table lookups are
required, which in theory makes this a much
faster solution
than more traditional IP
-
based
solutions


The MPLS header allows MPLS networks to be
shared among multiple customers, without
the customers seeing each other’s
traffic




Frame Relay follows the designed path of the
PVC to each site





From the telecom view MPLS uses switching
in the cloud to move the data to the next hop


For the customer view an IP routing protocol
provides a next hop for the destination IP
address


traditional Interior Gateway Protocols (IGPs)
like RIP, OSPF, and EIGRP do not work
well
over MPLS networks


BGP is the most common protocol in use to
communicate to customer
endpoints


So MPLS is then a Layer
-
3 protocol, not really


So it must be Layer
-
2 protocol, not
completely


It is like layer
-
2.5


If you could make Layer
-
2 that had routing
intelligence without the table overhead and
processing delay and fixed problems small
fixed cell size ATM adds you would be closer
to MPLS