Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

Multi Protocol Label Switching Recovery

Mechanism
Sulalah Qais Mirkar, Dr.Vijay Thakurdas
Raisinghani
School of Technology Management and
Engineering (MPSTME), Dept ofIT,
NMiMS Deemed-to-be University
sulalah.mirkar@nmim s.edu, rvijay@ieee.org
Abstract-A

circuit

to

sender

to

receiver.

next LSR, in the MPLS domain. A Label switched Path (LSP)

In case if a link on the path fails then data cannot be

is setup, through the LSRs, using a signaling protocol, such as

data

sent along the path.

switched

computer can

like voice and

video from

However circuit switching is resilient to

such failu res. These packets are treated individually in the


network. If failure occurs in the network path then packets can
be rerouted to avoid the failures and communication will not
be

Router (LSR). It labels the incoming packet and forwards it to

be used

send real time

interrup ted.

In

this

paper,

we

analyze

the

different

approaches of traffic rerouting in MPLS domain that are


resilient to the failure. We propose a new model Assured

Resource

Reservation
or

Protocol

(RSVP-TE)

[1][6]

Distribution

Protocol(CR-LDP)

with

Traffic

constraint-based
[1].

Engineering

Routing
label

Label

switched

path(LSP) within the MPLS domain decides the route of the


packet. A packet's label is removed by the Egress Router
(Egress LSR), before the packet is sent to non-MPLS router.

Quality-of-Service (ASQ) an improved mechanism such that

The working path or active path is the path which is

traffic demand can be rerouted in the network as fast as

reserved for transmission of traffic. To protect against failure

possible. Our approach helps to achieve a fast reroute of traffic

of working path, a backup path or recovery path is reserved.

on path failure, as compared to existing


Further, our

recovery models.

appro ach requires less num b er of nodes

on

the backup path as compared to recovery models proposed


by Makam's or Hask in.

Index

The recovery path [5] can be calculated and new setup


established when failure is detected.
There are two main [2] categories of recovery mechanism
depending on where a recovery can be placed.

Terms-Multi Protocol Label Switching,

Makam's

Global repair and Local repair. Global repair is

These are
to

protect

Model, Recovery Path Mechanism, Control-driven mode , Data

against any link or node failure on a path or segment of

driven mode.

the path. In global repair, the point of repair is usually distant

I. INTRODUCTION
Iti

Protocol

label

from the failure and needs to be notified by FIS. It is having an

Switching

(MPLS)

advantage that all links and nodes on the working path are

certain

that the FIS has to be propagated all the way back to the

[I]

is

technology which helps in Traffic Engineering.


S

[I]

is

technology

which

overcomes

limitations of IP based networks. In MPLS, the packets are


forwarded using short labels instead of the entire IP address. In
MPLS, since the forwarding of a packet is based only on label
switching, the packet forwarding is faster than IP network. In
MPLS, many approaches or models have been proposed to
move traffic from faulty active path to recovery path like
Makam's Model [2], Haskin Model [3] and Fast Reroute one
to-one back Model [4]. In Makam's Model, the working path
is disjoint with recovery path, which

leads to delay in the

Fault Indication Signal (FIS) reaching the upstream ingress


node. Haskin Model is also known

as

the Reverse Backup

Model. In this model, if a fault is identified in the working path


then traffic will be reversed from the faulty node to the ingress
node. So there is high probability that reverse traffic will
change the order of packets. The Fast route back-up Model
consumes

large amount of resources

like

bandwidth

and

routers. We observe that there are many challenges in existing


recovery models. Using our approach, we can overcome these
challenges.

protected by a single recovery path. But the disadvantage is


ingress LSR before recovery can start.
Local repair is to protect against a link failure or neighbor node
failure and to minimize the amount of time required for failure
propagation. In Local repair, a fault indication signal doesn't
have to propagate all the way back to the ingress router before
the recovery can start. Once the recovery is detected, recovery
can be start by the nearest LSP that detects that failure. Due to
this it is faster than global recovery. Local repair can be setup
in two different ways. i) Link Recovery:- The goal is to protect
a link in the working path from failures. ii) Node Recovery: The goal is to protect a node in the working path from the
failure. In Local protection, only one segment of the working
path is protected.
The rest of the paper is organized as follows. Section II
focuses on related work in MPLS recovery. In section III we
propose our recovery approach - ASQ. In section IV, we
compare existing models with ASQ model, using simulations.
Herein we also analyze the results. Section V concludes the
paper.

The entry point in an MPLS network is called the ingress

II. RELATED WORK

point [5]. This ingress router is called a Label Switching

978-1-4799-3140-8/14/$31.00 2014 IEEE

492

The resources required for backup path should be as low

There are two different ways of recovery , these are called


Recovery by Rerouting and Recovery Protection Switching [8].

and at the same time the recovery model should enable quick

i) Recovery by Rerouting:-It is a way of recovery in which a

switchover with low delay. Based on the above, we propose

recovery path is established on demand after a fault occurs

our recovery model below and discuss its salient features.

[6][8]. The Recovery path can be based upon fault information,


III. ASQ MODEL

network routing policies and network topology information.


ii) Recovery by Protection Switching:-In Protection switching
[7] a recovery path is pre-computed and pre-established before
a

failure

subtypes

occurs
of

on

the

protection

working

switching

path.

There are two

1+1 ("one

plus

one")

protection wherein resources like [2] (bandwidth, buffers and


processing capacity) on the recovery path are fully reserved.
This model is expensive and also recovery path is reserved
even when no failure occurs. The other type is

I: I ("one

for

one") protection wherein the resources [2] allocated on the


recovery path are fully available to low priority traffic except
when the recovery path is in use due to a fault on the working
path.

This model

is

less

expensive

as compare to

I+I

Protection switching model.


There are many approaches or models that have been proposed
to move the traffic from the faulty active path to recovery path.
In Makam's Model [2] the working path is disjoint with
recovery path. If a fault occurs on any link of the working path
then FIS needs to be sent to the path switch LSR or ingress
LSR to transfer traffic from the faulty working path to
recovery path.

In Makam's Model, a completely disjoint

We propose the ASQ model in which, a set of nodes


LSRs/links can be protected by a backup router.
General ASQ model: In general in the ASQ model, a backup

path is setup for the entire working path. The backup path is
disjoint

from

the

working path.

This backup path

is

connected with the working paths after every n hops. On


working path failure the nearest LSR which is connected to a
backup path, takes the switchover decision, the FIS is not
sent to the ingress LSR. The connection with backup path
after every n hops ensures redundancy and at the same time
lesser resources are needed. Further, since the nearest LSR
takes the switchover decision, the switchover is faster as
compared to other models.
An example implementation is shown in the Figure 1.
i)The backup path is setup using the LSRs 2, 6 and 10.
ii)The LSR

is the path switch LSR, responsible for

switching the traffic from active path to pre-established


backup path is connected with LSR 2.
iii) In this case, every alternate LSR, on the working path is

backup path is established in advance. This saves the time

connected with recovery path.

taken for computing the backup path when failures occur.

iv) Alternate LSR is only used as an example; a set of links

In

this model as the FIS has to travel from the failed node to
upstream ingress node (PSL), the delays in recovery initiation
are large. If sending rate is high from the source node than the
number of packets dropped are high.
In Haskin's Model [3], if a fault is identified in the current
working path then traffic will be reversed from the faulty
node to ingress upstream LSR. Reverse traffic works as a
Fault Indication Signal. So it reduces the time required for
the FIS to reach the ingress LSR. In this model, until the path

(more than two) could be protected by a single backup link.


In figure

I,

ASQ Model is shown with 1-9 MPLS nodes and

o and 11 are non MPLS nodes. In the next section, ASQ


recovery model is compared with the Makam model. This
comparison is done with 6-node, 12-node and 20-node chains
in control driven and data driven mode respectively. Control
driven and data driven modes are explained in the next
section.

switch LSR receives the reversed data packets, many packets


may already have been sent on the failed working path and if
transmission rate is high then number of packets dropped will
be high. There is high probability that the reverse traffic will
lead to a change in the order of packets, as compared to the
original traffic.
In Fast Reroute one-to-one backup model [4] every node in
the working path has its own separate backup, called a detour
LSP, which is computed in advance and reserved. Since each

LSR has a backup, whenever a fault occurs, the delay in


switchover is reduced since the FIS need not be sent to the
ingress LSR. This model consumes large amount of resources
like bandwidth and routers to connect with backup path.

Figure 1. ASQ Recovery Model

A. Simulation and Validation of ASQ model

In the above section, we analyzed various recovery models

We use ns2 for our simulations. We test our model through

for MPLS. We observed that the models have either long

6, 12 and 20 node chains in a working path. We assume a

delay in recovery (Makam's Model [2]) or require large

single intermediate link to connect to the backup path. Except

number of resources (Fast reroute model [4]) or cause packet

in the case of a 20 node link. The Fig. 2 is a nam file

reordering (Haskin's model [3]).


We defme the problem for our work as follows:

visualization in ns2 [9]. Nam is the tel based animation tool


that is used to visualize the ns simulation and real world
packet trace data.

2014

The nam file

contains

International Conference on Signal Propagation and Computer Technology (ICSPCT)

the topology

493

information like nodes, links, queues and node connectivity

are setup by sending mapping message from each LDP agent

etc. We implemented the ASQ model by editing the test

to its neighbors. We show below the results for control

suite-mpls.tel, available in tel examples file in ns2.

driven mode.

i) In ASQ recovery model, the node 0 is the source node and


Node

I I is the destination node.

ii) Node 1 is the ingress node or Path Switch LSR (PSL) and
node 2 is the egress Path Merge LSR (PML).

Control- driven Mode: 6- node chain, J 2-node chain and


20-node chain

a) 6-node chain
i) Link failure between LSR

I and 3

As shown in Figure 3, there is a link failure in the working

iii) The path 1-3-5-7-9 is the working path or the primary


path and the path 2-4-6-8-10 is the secondary path or backup
path.

path and backup path. So due to which two different fault


indication signals propagate in the network. One PIS will be
sent from the LSR 1 and 3 to its neighboring LSR's and

The working of the simulation setup, shown in figure 3, is as

another will be sent by the LSR 8 and 10 to their respective

follows.

neighbors.

Each MPLS Node will exchange LDP (Label

Distribution

Protocol)

mapping

request

sent

by

the

neighboring nodes. Each LSR will receive the LDP mapping


request. When Node 0 sends an IP packet to Node 9 in the
MPLS network, it sends an un-labeled packet (i.e. an IP
packet in an Ethernet frame without MPLS label). Node

is

the ingress LSR, after verifying the destination IP address

Now LSP between node 8 and 10 fails. So data packet will


travel through the nodes 1-2-4-6-8-7-9-11-13-15-16.

0-0070@@@@
-11=
-I
0G)0@-@)0)'-@

and other related information in the packet header, it pushes a

Figure 3. 6-chain control-driven ASQ Model in 1-3 link failure

label into the packet and forwards the labeled packet to the
output port. Node 3 LSR, receives the labeled packet from
the Node

LSR. It examines the label and performs a table

look-up at forwarding table to find a new label and the output


port. Node 3 then swaps the old label with new label and
routes the new labeled packet to the output port. Other LSRs
will perform similar tasks. The labeled packet will reach the
Node 9, the egress LSR. It then examines the label and
performs a table look-up at the forwarding table to find that
the packet is to be sent to non-MPLS Node
removes

the label

and

sends

then

I I. It

the unlabelled packet to

destination Node. When any link fails on the working path


(or backup

path),

the

downstream node

sends

fault

indication signal (FIS) to the nearest LSR connected to the


backup path (or working path).
In the ASQ Model, in the simulations, the center node of
backup path is connected to the working path using a link.

ii) Link failure between LSR 7 and 9


As shown in Figure 4, LSP fails between LSR 7 and 9. LSR
7 is directly connected to backup path. Therefore FIS will
reach to LSR 8 immediately. On receiving the FIS, LSR 8
will send the mapping message to its respective neighbors
and data packets will switch over from the working path to
recovery path. So the new path will be 1-3-5-7-8-10-12-1413.

o - 0.;,0-12J-0
f."\ '" r:;'\
f?I
(.;\
f.';\
@'
r.?I -I @
- t;;\
'-'V,I.!,!;13 .
I
;1
I
:
0-@-GY-@!.!.!@.!.!.@.!.!.@
...

...

..

16

Figure 4. 6-chain Control-driven ASQ model in 7-9 link failure

b) 12- node chain


i) Link failure between LSR 5 and 7

This will reduce the time required for the fault indication to

Similarly, in the 12-node chain shown in Figure 5, where link

reach the ingress LSR to notify about the failure in the

failure occurs between LSR 5 and 7, the LSR 5 sends the FIS

current working path.

to ingress LSR.

(';:\

f7\

I I

r:::'\

II I

f?\

t';\

\Q,J -\lI-0-0- 0
III
I
(D-G)-@-@
Figure 2. NAM trace file visualization of an ASQ Model with 3 node chain

We do a detailed comparison of both Makam's Model and


ASQ Model using different scenarios with 6-node, 12-node
and

20-node

chains.

While

comparison

considers

two

different modes of label distribution - one is control-driven


mode and another is data-driven mode.
1) The control-driven mode the LDP (label distribution

.........

.........

t::?\

I
@

Figure 5.12-chain Control-driven ASQ model after 5-7 link failure to backup
path

ii) Link failure between LSR 23 and 25


In Figure 6, the 12-chain ASQ model where failure occurs
between LSR 23 and LSR 25. LSR 23 will send FIS to
neighboring LSR. Once LSR 13 will receive the FIS, it will
immediately

transfer the control

from the current failed

working path to backup path.

protocol) distributes messages between all MPLS based


nodes even if there is no packet transfer taking place. LSPs

494

20 J 4

-tiJ;

International Conference on Signal Propagation and Computer Technology (ICSPCT)

@!!.!(!) .......... @-@)


1=
I;
1."::\
r.:?\
(D .......... -
.

...........

I
..........

@-@-' @
;1
t:":;\ t::?\

' -

Figure 6.12-chain Control-Driven ASQ model after 23-25 link failure

c)

Labels are removed from the packet

d)

Receive data packet at destination

I) Control-Driven Mode:
a)

Reception of first FIS at ingress router

As shown in Figure 9, the graph showing a reception of a


fault indication signal to a ingress LSR in 6-chain, 12-Chain

c) 20-node chain

and 20-chain.As seen from the fig.9, we can observe that

i) Link failure between LSR 5 and 7


In Figure 7, the 20-chain ASQ model where failure is occurs
between LSR 5 and LSR 7. LSR 5 will send FIS to

FIS takes less time to reach to ingress router in ASQ model


as compared to Makam's model.

neighboring LSR. Once LSR 1 will receive the FIS, it will


immediately

transfer the control from the current failed

working path to backup path.

'"
"
<:
0

06

..

0.4

0.3

..

0.1

0.5

0.2

.5

E
1=

ii) Link failure between LSR 35 and 37

12-Chain

6-Chain
0.350

20-Chain
0.479

0.412

0.320

0.371

0.372

Figure 9.Reception of FIS at ingress router in Controldriven mode

Similarly, Figure 8 shows the Link failure between LSR 35


and LSR 37. FIS is sent to LSR 21, which will switch over
the traffic from the current working path to recovery path.

I-Makam Model
I--Proposed Model

Figure 7. 20-chain Control-driven ASQ model after 5-7 link failure

---

b)

First label packet on backup path

Figure 10, shows the reception of first label data packet at the
backup path LSR, in case of working path failure. As shown in
the graph, a packet switches faster to backup path in the ASQ
model as compared to Makam's model.
o

0.6
<II

..,
c:

Figure 8. 20-chain Control-driven ASQ model after 35-37 link failure

2) In data-driven mode when data packets arrive, the label

binding are created by LDP (label distribution protocol)

.5
QI

between MPLS based nodes . The data packet contain source

E
F

and destination address. A multilayer switch constructed LSP


(label switch path) between nodes once it sees the first data
packet in traffic flows [10]. In this way LSP is constructed
from the source towards destination.
The comparison is done between Makam' s Model and

0.5
0.4
0.3

...

--=

0.2
0.1
a
6-Chain

-+-Makam Model

I -Proposed Model l

0.351
0.322

I
I
I

12-Chain
0.435
0.373

I
I
I

20-Chain
0.481
0.415

Figure IO.First Label packet on backup path in Control-driven mode

ASQ Model in 6-chain, 12-chain and 20-chain for both


control and data driven mode. Their graph and analysis is
shown in the next section. In the next section, we analyze

c)

Labels are removed from the packet

A packet's label is removed by the egress LSR, before the


packet is sent to the non-MPLS router. In fig.2, the LSR 9 is

the results of simulation.

an egress LSR wh ich removes the packet labels. The fig. I I


IV .RESULTS AND ANALYSIS
A.

shows that the data packet at egress router reaches early as


compared to Makam's model.

Results

The results show the comparison between Makam and ASQ


Models for the three different chain modes 6, 12 and 20 in
control

driven and

data

driven

mode. The simulation

analysis is done for different stages after a failure in the


working path. The stages are as follows:
a)

Reception of first FIS at ingress router

b)

First label packet on backup path

2014

International Conference on Signal Propagation and Computer Technology (ICSPCT)

495

0.8

"

<II

"C
0
u
OJ

0.6

0.4

<II
OJ

0.2

.5:

E
i=

I
I

....--:::

II)

"C

0.5

0
u

.,

6-Chain

-+-Makam Model

12-Chain

0.3

0.586

0.431

--Proposed Model

0.1

0.725

0.454

0.380

0.2

E
i=

20-Chain

0.543

6-Chain

i--Makam Model
I --- Proposed Model

Figure II. Labels are removed from the packet in Control-driven mode

Receive data packet at destination

...

0.4

.:
.,

d)

0.6

"

------

20-Chain

12-Chain
0.433

0.351

0.478

0.373

0.322

0.412

Figure 14.First Label packet on backup path in Data-driven mode

After removing a label from the packet, the egress router will
send the packet to the destination non-MPLS router. In fig.l2,

c)

Labels are removed from the packet

the packet reaches to the destination router in lesser time in

From Figure 15 too we can see that ASQ performs better as

ASQ model as compared to Makam's model.

compared to Makam's model.

0.9

'"

0.7

0.6

0.5

:E

0.4

.5
OJ
E
F

0.3

..-

0.6

0
u

0.5

.,II)
.

6-Chain

12-Chain

0.456

0.598

20-Chain

0.3

0.466

0.391

:
.,

0.554

0.1
0

6-Chain

i Makam Model
I - Proposed Model

Figure 12.Receive data packet at destination in Control-driven mode

---..-------

0.2

E
i=

0.841

....--

0.4

:E

--Proposed Model

0.432

.......

20-Chain

12-Chain

0.724

0.584

0.454
0.380
0.540
Figure 15 . Labels are removed from the packet in Data-driven mode

2) Data-Driven Mode:
a)

II)
"C

"

0.1

I....... Makam Model

.....

0.7

0.2
0

/"'"
-------

0.8

0.8

Reception of first FIS at ingress router

d)

Receive data packet at destination

As shown in Figure 13, the graph shows the reception of a

After removing a label from the packet, the egress router will

fault indication signal at

send the packet to the destination non-MPLS router. In Fig.16,

an

ingress LSR in data driven

mode. From the fig.13, we can observe that FIS is taking

the packet reaches to the destination router in lesser time in

less time to reach to ingress router

ASQ model as compared to Makam's model.

in ASQ model

as

compared to Makam's model.

"

<II

"C

0
u

0.4

0.3

OJ

E
i=

I
I

0.5

OJ

.5:

"
<II

"C

0.6

----

r-::::::

.5:

OJ

E
i=

0.2
0.1
0

6-Chain

12-Chain

20-Chain

--Makam Model

0.350

0.412

0.476

---Proposed Model

0.320

0.351

0.372

Figure 13.Reception of FIS at ingress router in Data-driven mode

b)

0
OJ

First label packet on backup path

Figure 14 shows the reception of first label data packet to


backup path LSR in case of working path failure. As shown in
the graph, a packet switches faster to backup path in ASQ
model as compared to Makam's model.

I
I

0.8
0.7
0.6

0.5
0.4

--

--

0.3
0.2
0.1
0

6-Chain

.......Makam Model

0.456

--Proposed Model

0.392

20-Chain

12-Chain
0.596
0.466

0.735
0.551

FIgure l.Recelve data packet at destmatlOn 10 Data-dnven mode

B. Analysis of Results
The ASQ Model performs better in both control driven and
data driven mode as compared to Makam's Model as given
below.
i) Control Driven mode
We see the best performance of ASQ in the 20 chain case. As
the number of network nodes increases in, the Makkam

496

20 J 4

International Conference on Signal Propagation and Computer Technology (ICSPCT)

model requires more time to switch from faulty working path

network. If Quality of Service requirement is low, then it can

to predefined backup path. The first FIS takes 28% more time

be better option as it is less expensive as compared to the

to reach at ingress node, due to which packet transfer is

other recovery models in MPLS network. The ASQ Model is

delayed by 15% and also packets from recovery path to

a better option if Quality of Service is of high priority.

destination is delayed by 51%, as compared to the ASQ

However, it will be expensive as compared to Makam's

model. For other cases also ASQ shows better performance

model since multiple additional links are required to connect

than

to

Makam.

Table

below

summarizes

the

results

the

backup

path.

Our simulation results comparing

comparing ASQ Model and Makam Model, for control

Makam's Model and our ASQ Model show that in a huge

driven mode.

network where number of node connections are more, the


TABLE I

COMPARISON SUMMARY FOR20-CHAlN, CONTROL DRIVEN MODE - ASQ VS


MAKAM
No.

Metric for
comparison

ASQ Model in
milliseconds

Makam Model in
milliseconds

I.

Early
notification to
ingress node

0.372 ms

0. 479 ms

2.

Switching time

0.415 ms

0. 481 ms

3.

Reception of
data at
destination

0.554 ms

O. 84l ms

ASQ model shows better performance. This is because the


time taken by the Fault Indication Signal to reach the Path
Switch LSR is higher. This time will increase as more
number of nodes and links will be added and therefore more
delays will exist in case of Makam's model. In our ASQ
Model, the above problem has been addressed and the time
taken by the Fault Indication Signal to reach the ingress
router has been reduced, due to multiple connections between
working and backup path.
This also resu Its in the packets reaching faster to its destination
as compared to Makam's Model, after a fault has occurred on
the working path. Further evaluations are needed to compare

ii) Data Driven mode


Similarly, in the 20-node chain Makam Model,

FIS takes

28% more time to reach the ingress router; packet transfer on

the ASQ model with other recovery models and to evaluate the
behavior with different traffic conditions.

recovery path will be delayed by 16% and first data packet


will take 33% more time to reach at destination node, as
compared to ASQ model. Similarly, in 6 and 12 node chains,
ASQ shows better performance. Table II below summarizes

REFERENCES
[I]

the results comparing ASQ Model and Makam Model, for


data-driven mode.

[2]

The reason that ASQ performs better is that ASQ provides a

[3]

connection to the backup path from the working path at


intermediate nodes. This helps in quick switchover to the
backup path on any link failure. The Table II is given below
which shows comparison between ASQ Model and Makam
Model in 20-chain data driven mode.

[4]
[5]
[6]

TABLE II

COMPARISON SUMMARY FOR 20-CHAlN, DATA-DRNEN MODE- ASQ VS

[7]

MAKAM
Metric for
comparison

ASQ Model
observation in
milliseconds

Makam Model
observation in
milliseconds

I.

Early
notification to
ingress node

0.372 ms

0. 476 ms

2.

Switching time

0. 412 ms

0.478 ms

3.

Reception of
data at
destination

0.551 ms

0. 735 ms

No.

[8]
[9]
[10]

Changcheng Huang, "Building Reliable MPLS Networks using a path


protection Mechanism", Carlton University: IEEE Communications
Magazine, pp.156-162,2002.
S.Makam's, V. Sharma, K.Ownes, C Huang "ProtectionlRestoration
Of MPLS Networks" draft-Makam's-MPLS-protection-OO.txt,1999.
D.Haskin, RKrishnan, "A Method of setting an alternative Label
switched path to handle Fast Reroute", draft-haskin-MPLS-fast
reroute-OS. txt, 2000.
K. Kompella, G.Swallow, "Detecting MPLS Data Plane Failures", Draft
ietf-mpls-lsp-ping-06.txt, July.2004.
Petersson,J.M, "based Recovery Mechanism", University of Oslo Master
Thesis, 2005.
S.Yoon, H. Lee, D. Choi ,Y. Kim "An Efficient Recovery Mechanism
for MPLS-based Protection LSP" IEEE ICA TM-200I September
2001.
V. Sharma, F. Hellstrand "Framework for Multi-Protocol Label (MPLS)
based Recovery" IETF,RFC 3469 February 2003.
S. Veni and Dr. G. M Kadhar Nawaz,"Protection Switching and
Rerouting in MPLS", India :[EEE Conference ,pp.216-220,201 O.
VINT project at LBL, Xerox PARC,USB and USC/IS The Network
Simulator NS2 htm://www.isi. edulnsnam/ns/ (accessed on 2013)
Chuck Semeria. (1999, Sep 27). Juniper Networks [Online].
Available:http://mirror. unpad. ac. id/ orarillibrary/library-ref-eng/
ref-eng- 3/network/mpls/20000I. pdf

V. CONCLUSIONS AND FUTURE


WORK
The Makam's Model has its own disadvantages and
advantages. It usage depends on the requirements in the

2014

International Conference on Signal Propagation and Computer Technology (ICSPCT)

497

You might also like