ason-wson fundamentals rev9

8/10/2019 ASON-WSON Fundamentals Rev9

1/56

1

ASON/WSON Fundamentals

Welcome to the ASON / WSON Fundamentals e-Learning Course.


2/56

2

Objectives

During this course we are going to focus on:

An introduction toASON/WSON technology

A description of theASON/WSON Control Plane

A description of theASON/WSON Protections

On completion of this course the participants will be able to:

1.Describe whatASON/WSON is and which are its main characteristics.

2.Differentiate betweenASON and WSON

3.Understand how WSON works and the relevant characteristics in terms of control plane and

protection schemes introduced into the WDM layer

During this course we are going to focus on:

An introduction to the ASON/WSON technology

A description of the ASON/WSON Control Plane

A description of the ASON / WSON Protections

On completion of this course the participants will be able to:

Describe what ASON is and which are its main characteristics.

Differentiate between ASON and WSON

Understand how WSON works and the relevant characteristics in terms of control plane

and protection schemes introduced into the WDM layer


3/56

3

Introduction

Lets introduce ASON / WSON main general concepts.


4/56

4

WDMSDH

Terminology

ASTN(ITU-T)

WSON(IETF)

ASON(ITU-T)

GMPLS : Generalized Multi-Protocol Label Switching (IETF)

ASON : Automatic Switched Optical Network (ITU-T / General Optical Domain)

MPLS TP : Multi-Protocol Label Switching Transport Profile

OTN (ODU Switching): Optical Transport Network (Optical Data Unit Switching)

ASTN : Automatic Switched Transport Network (ITU-T / SDH Domain)

WSON : Wavelength Switched Optical Network (IETF / WDM Domain)

Optical Domain

SDH Domain WDM Domain

GMPLS

(IETF)

MPLS-TP

(ITU-T / IETF)

OTN

(ODU Switching)

(IETF)

ASON, ASTN, WSON, GMPLS

Before beginning with the details about the ASON/WSON technology and

according to the fact that its not unusual to listen to people speaking about

ASON, ASTN, WSON, GMPLS, as if all these technologies were the same, itseems to be useful to clarify some points about the terminology.

The acronym GMPLS, that stands for Generalized Multi-Protocol Label

Switching, refers to a suite of protocols developed by IETF to extend the MPLS

ideas outside the context of the IP world.

The acronym ASON, that stands for Automatic Switched Optical Network, is a

Recommendation developed by ITU-T that specifies the requirement to apply the

GMPLS technology to a generic optical network; in this context, with the term

generic optical network we refer both to an SDH or WDM network, or even to

OTN (ODU switching) an MPLS-TP packet network.

In the SDH domain, the specific used acronym is ASTN, that stands forAutomatic Switched Transport Network, and, also in this case, it is a

Recommendation developed by ITU-T. Its a framework that represents the ITU-T

ideas about how the GMPLS technology should be applied to the SDH world (fast

rerouting of Virtual Containers).

In the WDM domain, instead, the specific used acronym is WSON, that stands for

Wavelength Switched Optical Network, and this a draft, not yet a

Recommendation, developed by IETF, that represent the IETF and ITU-T vision

of how the GMPLS should be applied to the optical part of the WDM world (fast

rerouting of wavelength).


5/56


6/56

6

WSON evolution - advantages

WSON

Advantages

Advanced NetworkResilience

Automatic Circuit

Provisioning

Reduced Costs

Why do we need WSON?

What are the main advantages of the WSON network compared with a traditional

network?There are three main advantages:

The most important advantage is that, in WSON, new protection schemas allow

advanced network resilience mechanisms, that can react, in an automatic

way, also in case of multiple failures.

The second advantage is the possibility for WSON to realize the automatic circuit

provisioning: as a strategic point of evolution, WSON will be able, in the

future, to provide circuits on demand, allowing an IP router, for example, to

ask for a circuit directly to the WDM node that is at the ingress of the WSON,

without passing through the NMS and the human operator.The third advantage is the strategic cost reduction mainly due to the fact that it is

possible to share the protection bandwidth among a group of WSON circuits.


7/56

7

a Traditional Network

Ethernet Node

WDM Node

Network Management System (NMS)

Ethernet Node

Data Communication Network (DCN)

To implement these advantages, WSON is quite different compared with a

traditional WDM network.

To understand what are these differences, lets take a look at what happens in atraditional WDM Network.

Lets consider the WDM network in the picture: this network has 6 interconnected

WDM nodes.

In a traditional WDM network the nodes are not aware of the fact that they are

part of a network.

For example, the nodes dont know the topology of the network and are not able

to take any decision about where to put the customer circuits; even the entity

circuit has no meaning for a single WDM node; in other words, the nodes are not

network aware.Who has the ability to understand how the nodes are interconnected? Who has

the intelligence to understand that a list of cross-connections between two

termination points is a circuit? In other words, who is the brain of the network?

In a non WSON environment the intelligence of the network only resides inside

the NMS and the human operator that sits in front of it.

The decision taken by NMS and the human operator are sent individually to each

node by the DCN.

The nodes, passively execute the commands coming from the NMS; they dont

take part in any routing decision; they are not able to interact with the other nodes

in order to create or delete circuits.


8/56

8

Ethernet Node

WDM Node


Ethernet Node


a wson Network

In a WSON network the situation is different: the intelligence of the network is

now distributed among the nodes, the NMS and the human operator.

Specific WSON processes run inside the controller cards of each node, thatimplement, together with the WSON processes running inside the NMS, the

distributed Control Plane, that is the distributed intelligence that manages the

customer traffic.


9/56

9

Evolution: from a centralized to a

distributed control plane

CENTRALIZED

CONTROL PLANE

DITRIBUTED

CONTROL PLANE

WSON

So the first point of distinction between a traditional WDM network and WSON is

the evolution from a centralized intelligence to a distributed intelligence, therefore

the evolution from a centralized control plane to a distributed control plane.


10/56

10

Evolution: from a centralized to a

distributed control plane

DATA PLANE

DISTRIBUTED CONTROL PLANE

The distributed control plane controls the data plane, that is the logical entity

formed by all the resources available in the network to transport customer traffic.

The data plane transports the customer traffic; the control plane decides wherethis traffic should pass through and sends the right commands to the involved

data plane resources.

The distinction between Control Plane and Data Plane is important in a WSON

context.

Please notice that, the fact the control plane is distributed among different

entities, means that these entities must be coordinated: the complexity of the

WSON Control Plane is mainly due to its distributed nature.


11/56

11

WSON internodes Communication network

Ethernet Node

WDM Node


Ethernet Node


WSON Inter-node

Communication

Network

In order to be able to implement a distributed intelligence, the WSON equipments

must have , first of all, the capability to speak with each other, not only with the

NMS.This is obtained by an inter-node communication network that is realized using

specific WSON reserved communication channels; in GMPLS terminology, these

special channels are called Control Channels.

Its important to remark that, this new intra-nodes communication network doesnt

substitute the traditional Data Communication Network, but works together with it.

The DCN must be present in any case, to allow the communications between the

nodes and the NMS for both WSON and non WSON related operations. For

example, the performance data collection is realized using the traditional DCN;

the control channels are not involved in this process.


12/56

12


Logical Control Channel

that runs inside the fiber

together with the customer traffic

IN-FIBER-IN-BAND

IN-FIBER-OUT-OF-BAND

Depending on the physical implementation its possible to classify the control

channels in different categories.

An in-fiber-in-band control channel is a logical communication channel thatshares the bandwidth with the customer traffic and that runs in a fiber. The

bandwidth available for the communication is taken by the bandwidth available for

the customer traffic; if there is no bandwidth consumption for the communication,

it means that more bandwidth is available for the customer traffic. Typically, the

routers in the IP world use in-fiber-in-band communication: control information is

sent together with the customer traffic.

In in-fiber-out-of-band control channels, the bandwidth reserved for the traffic and

the bandwidth reserved for the communication are completely not overlapping.


13/56

13


LOGICAL CONTROL CHANNELS IMPLEMENTED

USING THE ETHERNET Q INTERFACE

(NORMALLY RESERVED FOR THE COMMUNICATION WITH THE NMS)

OUT-OF-FIBER-OUT-OF-BAND

Ethernet INTERFACE Ethernet INTERFACE

The other type of control channel is the out-of-fiber-out-of-band control channel:

the typical example is when we use the Ethernet interface, that is normally

reserved for the communication with NMS, to transport the logical controlchannel; in this case, the channel is out of fiber, because it runs outside the fiber

transporting the customer traffic and it is out of band, because the bandwidth on

the Ethernet interface is reserved for the communication and cannot be used to

transport traffic.


14/56

14


123

80

.

..

Optical Supervisory Channel (OSC)

Bandwidth reserved for the in-fiber-out-of-band control channel

As far as DWDM systems are concerned, in particular, the relevant standards

define an additional low rate lambda that is specifically designed to transport

management information.The logical channel that uses this special lambda is called Optical Supervisory

Channel or OSC.

Inside the OSC, a configurable amount of bandwidth can be reserved for WSON

internodes communication.

This type of control channel is in-fiber, because it shares the physical fiber with

the customer traffic, and it is also out-of-band, because the amount of bandwidth

reserved for the control channel inside the OSC cant be used to transport any

other customer traffic.

In special applications, like the single hop long distance, this OSC can also beimplemented on one of the wavelength part of the standard grid for working

traffic, loosing therefore a traffic channel, but achieving longer distance coverage

for the OSC itself.


15/56

15

Requirements For a WSON

What are

the general

requirements

for a WSON

?

The nodes must be able to exchange

information about the status of the network

The nodes must be able to send commands

each other

The nodes must be able to re-route the traffic

from one line to another line even changing the

lambda of the circuit

But, what are the general requirements to realize a WSON able to take

advantage of the advanced new protection schemas?

A WSON is based on the following assumptions:The nodes are able to exchange information about the status of the network;

The nodes are able to send commands to each other;

The nodes are able to re-route the traffic from one line to another line even

changing the lambda of the circuit.


16/56

16


LINE 1

LINE 2

LINE3

LINE

5

LINE6

LINE 7

LINE8

LINE 4

LINE 9

CUSTOMER

TRAFFIC

ROADM

A WDM node that is able to reconfigure the direction of the traffic via software

commands is called multi-directional Reconfigurable Optical Add and Drop

Multiplexer or ROADM: all WSON capable nodes are multi-directional ROADMs.But how is the capability to change the lambda used by a circuit using only one

transponder obtained?

This is obtained using a special kind of transponder that is called tunable

transponder: this transponder can change the lambda of the circuit changing its

internal configuration via software.

Please notice that, its always the controller of the WSON node, guided by the

distributed control plane processes, that triggers the change of direction and, if

necessary, lambda for a circuit.


17/56

17


WSON NODENMS

WHERE TO RE-ROUTE

THE TRAFFIC ?

OR

THE DECISION CAN BE TAKEN

BY

PCE

PCE

One of the WSON requirements is that the WSON nodes should be able to re-

route traffic when a failure occurs.

In order to be able to create or re-route WSON circuits, the equipment must knowwhich is the best path to choose (click);

How is this knowledge obtained?

We can have two alternative approaches:

The decision is taken by some process running inside the NMS; The NMS tells to

the nodes which is the best path to create;

The decision is taken by the WSON node itself; its able to calculate where to

route o re-route the traffic without passing through the NMS.

In both cases, the engine that is in charge to calculate the best worker and all the

best protections path, according to some specific constraints, that its possibleto explicitly define, is generically called Path Computation Engine or PCE.


18/56

18

The Control Plane

Lets start describing in deeper details the main features of a WSON

Control Plane.


19/56

19

Ideal situation

1. INFORMATION ABOUT

THE NETWORK TOPOLOGY:

GOSPF-TE

2. CALCULATION OF THE BEST PATHS: PCE

3. SENDS COMMANDS TO OTHER NODES

TO CREATE AND PROTECT THE CIRCUITS:

GRSVP-TE

WSON NODE

Lets now consider an ideal situation for a while.

The ideal situation is the one in which the PCE runs inside each single WSON

node: this would allow each node to take real-time decision very fast,independently from other nodes and from the NMS. As a consequence, more

efficient protection schemas could be implemented.

To make it work, the first step is to allow the WSON nodes to know the topology

of the entire WSON network.

This information can be acquired using one of the GMPLS supporting protocols,

the GOSPF-TE.

GOSPF-TE is a routing protocol and its purpose is to allow WSON nodes to

exchange information about the network topology. This information is: the

administrative costs of the links all around the network, that are assigned viaconfiguration; the bandwidth available or the bandwidth used for each link of the

network; the status of each link: is the link up or is it down?; other additional

administrative information that could have been assigned to the links for different

purposes.


20/56

20

Ideal situation



GOSPF-TE

2. CALCULATION OF THE BEST PATHS: PCE



GRSVP-TE

WSON NODE

The second step is to calculate the best path: having the picture of the network,

every WSON node is able to find out what is the best path to follow to create or to

protect a circuit; the tool that is in charge of calculating best paths is the PCE.


21/56

21

Ideal situation



GOSPF-TE

2. CALCULATION OF THE BEST PATHS:

PCE



GRSVP-TE

WSON NODE

Step number 3 is the actual creation or re-routing of a WSON circuit: once

decided the best path, a WSON node will send requests to the other nodes on

the selected path in order to create, re-route or release WSON circuits; this isdone using another GMPLS supporting protocol: the GRSVP-TE.

GRSVP-TE is a signaling protocol and its purpose is to allow WSON nodes to

send commands, and receive feedbacks from the other nodes, in order to create,

re-route or delete WSON circuits.

Notice that WSON nodes always exchange topology information, using the

GOSPF-TE, even if there are no active circuits in the network;

On the other side, the signaling protocol, the GRSVP-TE, is active only at the

moment when a circuit is created, re-routed or deleted and only between the

involved nodes.


22/56

22

THE Path computation engine

PCE

CALCULATE

THE BEST PATHS

PERFORM

THE RWA

VALIDATE

THE PATHS

Probably, the path computation engine is the most critical component of a

WSON.

In general the functions of a PCE are:Calculating the best paths;

Performing the Routing and Wavelength assignment;

Validating the paths.


23/56

23


PCE

CALCULATE

THE BEST PATHS

PERFORM

THE RWA

VALIDATE

THE PATHS

In a WSON, the work of the PCE is complicated by the fact that calculating thebest path in the photonic world is not as easy as it would be in SDH world.

In the photonic world there are some additional physical constraints to be takeninto consideration, such as Polarization Mode Dispersion, non-linear effects, notperfect signal amplification and so on; these constraints have a more criticalimpact in WDM world compared with SDH world.

For example, in the photonic world, the fact that its possible to create a circuitbetween a node A and a node B and that its possible to create a circuit betweennode B and a node C, doesnt imply that its always possible to create a circuitbetween node A and node C.

Another critical point is that, in WDM world, given the fact that it is not possible tochange the frequency of a circuit without going in the electrical domain, itsnecessary to try to minimize the usage of the lambdas all around the network; the

problem to find out the best path in terms of administrative cost, minimizing, atthe same time, the number of lambdas used in the network is known in literatureas the Routing and Wavelength Assignment problem (RWA). The RWA has beendemonstrated to be an hard problem to solve; practically, a real PCE will solvethe problem using approximated solutions, based on heuristic, that aretechniques designed to solve a problem that ignores whether the solution can beproven to be correct, but which usually produces a good solution.

In conclusion, the work of the PCE is to resolve, for each circuit that itsnecessary to create o protect, the RWA problem, taking into account all thepossible physical impairments specific for the network: every path must becalculated and validated, checking if the path is physically feasible despite all the

physical impairments.


24/56

24


PCE

CALCULATE

THE BEST PATHS

PERFORM

THE RWA

VALIDATE

THE PATHS

Its not easy to realize an efficient engine that is able to perform all these tasks:

especially the validation step can be a long process in a big network.

This is the reason why its difficult to have PCE running inside the single WSONnode.

More commonly, WSON nodes interact with a PCE that works off-line, to have

the time to plan all the paths performing calculation, RWA and validation.

The circuits that pre-calculate and pre-validate the protection paths before a

failure happens are called pre-planned protected circuits.

In a first phase, all the WSON protections are pre-planned;

In a second phase, when the PCEs will run directly inside the WSON nodes

controllers, other kinds of protection will be available, like the so called on the fly

protections, in which the protection path is calculated and validated on the fly,real-time, only when a failure occurs.

Notice that, even in case in which the PCE works off-line, the routing protocol is

still required to be running to monitor the real time status of the network

resources; otherwise, the nodes would never know if a link that is part of a pre-

calculated protection path is still available or not.


25/56

25

acquiring topology information

Control Plane configuration: costs assignment

COST 10 COST 20

COST 5 COST 15 COST 20

COST 30COST 10

When setting-up a WSON, its always necessary to configure some parameters

that will be used by the GMPLS processes running in each WSON node; this

phase is known as configuration of the control plane parameters.During the configuration of the control plane, one of the steps is to assign some

administrative constraint to each link on the network.

This operation can be done manually, or using some automatic tool that can help

to avoid mistakes.

Once configured, these parameters will be used by the GMPLS routing protocol,

the GOSPF-TE.

Now, lets see and example of how WSON nodes acquire information about the

topology of the network, that is the first step to calculate the best path.

One of a parameter to assign to the links is the administrative cost; it representsthe cost to send traffic out of the interface connected to that link.

This information will be used by the PCE, because it will try to minimize the total

cost of a path represented by the sum of the cost of the links it passes through.


26/56

26

acquiring topology information

COST 10 COST 20


COST 30COST 10

1 2 3

4 5 6

LINK 1 LINK 2

LINK 3

LINK 4

LINK 5

LINK 6

LINK 7

LSULSU

LSU

LSU

LSU

LSU

LSU

Im the node 1

Im connected to the node 2

through link 1,

the cost of which is 10;

Im connected to the node 4

through link 7,

the cost of which is 5

Information

packet

LSU

How does it work?

Every node in the WSON is responsible for sending detailed information about its directly

connected link to all the other nodes in the network.

Consider node 1 in the picture: when the GOSPF-TE process is enabled, after an initial

handshake to its adjacent neighbors, the node 1 builds a packet containing at least the following

information in order to define some common parameters: Im the node 1, Im directly connected

with node 2 through the link 1, that has the cost of 10 and Im connected to the node 4, through

the link 7, the cost of which is 5.

In the context of GOSPF-TE, this information packet is called LSU, Link State Update and the

piece of information describing each link is called Link State Advertisement or LSA.

Once this information packet is created, its sent out to all interfaces of node 1.

When the node 2 and the node 4 receive the packet coming form node 1, they store the

information inside a local topology database and send an updated copy of the packet to all their

own adjacent neighbors, but not the one from which the information packet has been received.

When the node 3 and the node 5 receive the packet coming from node 2 and node 4, respectively,

they store the information inside a local topology database and send an updated copy of the

packet to all their own adjacent neighbors, but not the one from which the information packet has

been received.

Finally, also the node 6 receives the information coming from the node 1 and stores this

information inside its local database.

The final result is that the piece of network described by node 1, regarding how link 1 and link 7

are interconnected to the other nodes, is known by all the nodes in the network.

In a similar way, also the other nodes in the network create an information packet regarding their

own directly attached links, and flood it to all the other nodes in the network.

At the end of this flooding process, every node will have exactly the same picture of the network.

On the base of this picture, the PCE will apply its algorithm to calculate the best paths.


27/56


28/56

28

GOSPF-TE: TE information

Additional Traffic Engineering information:

In GMPLS context, with the term Traffic Engineering, we refer to the capability to

take routing decision based on some additional information that is not possible to

express only with a fixed cost given to a link.What we have seen till now, regards only the part of the GOSPF-TE that is not

related to the traffic engineering functionalities.

To support the traffic engineering functionalities, the LSU packet must transport

some additional information regarding the links, not only the administrative cost.

For example, we can think of assigning an higher cost to links that have a very

high percentage of used bandwidth, so we can imagine a dynamic cost

associated to the link that can vary with the bandwidth occupancy.

Every time that there is a variation on the bandwidth, for example, when a new

circuit is created or is released, new LSU are sent by the nodes that are adjacentto the links involved in the paths to all the other nodes.

This functionality is difficult to use when the PCE works off-line, because all the

path calculation is done in advance and so the dynamic cost cannot be taken in

account.

This functionality is important when the PCE run directly inside the nodes and

can calculate and validate the paths real time.

Notice that, even if the PCE works off-line the status of the network must be

monitored because the WSON nodes must always know which of the links are

available and which are not.


29/56

29



Total Reserve-able Bandwidth

Bandwidth Available

The total reserve-able bandwidth: for each link, how many channels can I reserve

for WSON paths? All? Only a part? Is this an 80 channels link o a 40 channels

link?The Bandwidth available: how many lambda are free in a moment to create new

paths?


30/56

30



Colors

Shared Risk Link Groups (SRLGs)

Two optional additional information can be transported by the LSU:

Colors


A color is an attribute that can be assigned to a link, so to distinguish a group of

links from the others.

For example, a red color can be assigned to a group of military links: when we

create a circuit, it possible to request that the circuit pass only through the links

that have the red color as attribute configured during the control plane

configuration phase. If its not possible to satisfy the color constraint, the circuit in

not created at all. Among the links respecting the color constraint, the one with

the lowest cost path is selected.

Notice that the color constraint, not only is stronger than the administrative costconstraint, but it is also blocking: if there is no way to connect the two termination

point of a circuit, the circuit is not created at all and a message is sent to the

operator.


31/56

31



Colors


The concept of Shared Risk Link Group in used by the PCE when calculating the

protection path for a circuit. It represents a group of links that, somehow, share

some risk. For example two links passing through the same physical conduit insome point share a risk, in the sense that, if someone damages the conduits,

probably the two links will be broken at the same time. During the configuration of

the control plane is possible to specify the SRLGs of which a link is part of: its an

additional topological information that cant be given only with the administrative

cost metric.

When are SRLGs used? An operator can specify that he would like to have the

worker and protection paths passing through completely disjoint SRLGS. If, for

example, the worker path pass through links that are part of SRLG 1, SRLG 2

and SRLG 3, the PCE will try to avoid the usage of links that are parte of SRLG1,

SRLG 2 or SRLG 3.As it was for the colors constraint, the SRLGs constraint is stronger than the

Administrative cost. But, unlike the case of the colors constraint, the SRLG

constraint is not blocking: even if it is not possible to respect the SRLGs diversity

between worker and protection paths, the circuit is protected, if a path is

available.


32/56


33/56

33

GOSPF-TE: colors

Control Plane configuration: colors assignment

COST 10 COST 20


COST 30COST 10

COLOR RED

COLOR RED

COLORS RED

AND GREEN

LINK 1 LINK 2

LINK 3

LINK 6

LINK 7LINK 5

LINK 4

COLOR GREEN

NODE 1

NODE 4

Failure on a link

NO PROTECTION PATH AVAILABLE: THE CIRCUIT IS NOT PROTECTED

The PCE, before calculating the best path, deletes from the picture of the network

the links not respecting the color red constraint: link 2, link, 3, link 6 and link 7 are

deleted from the topology.Now the PCE calculate the best path taking into account the administrative costs;

there is only one path available: the one passing through link 1, link 4 and link 5;

node 1 sends commands to the nodes involved in the path and the circuit is

created.

Notice that, once a color attribute has been assigned to the links, it can be used

or not; if the operator doesnt specify a color during a creation of the circuit, the

color attribute is not taken into consideration by the PCE.

Another point is that, in case the operator wanted this circuit to be protected by

another path passing only through red links, if there is a failure on one of the links

of the working side, its not possible to protect the circuit anymore, because thereis no other path that can join node 1 and node 4 passing only through red links.


34/56

34

GOSPF-TE: srlgs

COST 5 COST 15

COST 30COST 10

COST 10/SRLG 1/LINK 1

SRLG 3

LINK 7




COST 20SRLG 4

LINK 5

SRLG 5

LINK 9

SRLG 6/ LINK 8

Control Plane configuration: SRLGs assignment

SRLG 7/ LINK 6

Lets now see an example for SRLGs.

As we said, it can be used to inform the PCE that two links pass through the

same physical conduit in a point of the network; this information can be useful,because we can suppose that, probably, if there is a damage on the conduit, both

the links will go down.

When is this information used?

An operator can request to the PCE to satisfy an SRLG diversity between worker

and protection path during the creation of a circuit.


35/56

35

GOSPF-TE: srlgs

COST 5 COST 15

COST 30COST 10


SRLG 3

LINK 7




COST 20SRLG 4

LINK 5

SRLG 5

LINK 9

SRLG 6/ LINK 8

Control Plane configuration: SRLGs assignment

SRLG 7/ LINK 6

Failure on a linkTHE CIRCUIT IS PROTECTED TRYING

TO RESPECT THE SRLG DIVERSITY

Failure on a second link

THE CIRCUIT IS PROTECTED EVEN IF

IT IS NOT POSSIBLE TO RESPECT

THE SRLG DIVERSITY

Lets take a look at the picture;

Suppose the worker path of a circuit pass through links 1, 3 and 5, that are part of

SRLG 1, SRLG 2 and SRLG 4, respectively. If the operator specifies that hewants to have SRLG diversity between worker and protection, this means that the

PCE will try to avoid the usage, for the protection path, of links that are part of

SRLG 1, 2 and 4.

This is done by the PCE, assigning a very high cost to links not respecting the

diversity during the calculation of the best path.

Anyway, this constraint is not blocking, like it would have been with colors: if

there is a path available, even not respecting the diversity, the circuit is protected

in any case; in this situation, the cost of the protection path will be very high.

Lets see an example: If there is a failure on link 3, the traffic is switched on theprotection path passing through the SRLG 5, 6 and 7: the SRLG diversity

requirement is respected.

Now, suppose there is a second failure on the link 6. The traffic is switched on a

second protection path passing through SRLG 1, 2 and 4: the SRLG diversity

requirement cant be respected but, anyway, the traffic is protected.


36/56

36

GRSVP-TE: signaling a path

OFF-LINE

PCEWSON

VALIDATED

LSPs

CIRCUIT REQUEST

NMSCIRCUIT REQUEST

WITH PATH SPECIFICATION

Once the PCE has calculated and validated a circuit, worker and protection

paths, the next step will be to setup the circuits, that, in a WSON, must always be

bidirectional.The validates paths are therefore sent by the PCE to the NMS, that passes these

information on to the ingress node for the specific circuit.


37/56

37


LINK 1 LINK 2

LINK 7 LINK 5 LINK 3

LINK 4LINK 6

NODE 1

NODE 4

NODE 2 NODE 3

NODE 5NODE 6

Control Channel Control Channel

ControlChannel

PCE

The management of WSON circuits is delegated to the GMPLS signaling

protocol: the GRSVP-TE.

GRSVP-TE is used to setup and tear down circuits.A circuit is the entity that is formed by both worker and protection paths.

In GMPLS terminology, a path is called Label Switched Path or LSP, because,

during the setup, the nodes exchange messages about resource reservations

identified by labels.

A circuit is formed by one or more LSPs.

But what is a label? A GMPLS label represent a traffic resource on a link: there is

one label for each lambda that is possible to reserve in a link; for example, if ,

after the Routing and Wavelength Assignment phase, the PCE has decided to

reserve the first channel of the eighty available for a circuit, it indicates thischannel with a label, that represents the channel 1.

The information about the chosen label is transported inside GRSVP-TE

messages.


38/56

38


LINK 1 LINK 2


LINK 4LINK 6

NODE 1

NODE 4

NODE 2 NODE 3

NODE 5NODE 6


ControlChannel

PCE

Now, lets consider an example: an operator wants to setup a protected circuit

between node 1 and node 4; he inserts his request inside the NMS.

At this point two things can happens:1. if the PCE process runs inside the nodes, the NMS sends its request of circuit

creation to the node 1, that represents the ingress node for the circuit; the

PCE calculates the best path for both worker and protection side, performs

Routing and Wavelength Assignment and validates the traffic; the output of

the PCE is a list of nodes, links and lambdas for both worker and protection

side, that will be used by the GRSP-TE to signal the circuit.

2. if the PCE runs off-line, not inside the nodes, the PCE calculates the best path

for both worker and protection side, performs Routing and Wavelength

Assignment and validates the traffic. The output of the PCE is sent to the

ingress node, and the GRSVP-TE process will use this information to signalthe circuit.


39/56

39


LINK 1 LINK 2


LINK 4LINK 6

NODE 1

NODE 4

NODE 2 NODE 3

NODE 5NODE 6


ControlChannel

PCE

PATH PATH

PATH

RSV

RSVRSV

The circuit is created: the status of the circuit is transmitted to the NMS

Please, notice that, in both cases, the signaling phase is performed directly by

the ingress node, not by the NMS: in the first case, we can think about a

centralized routing and a distributed signaling behavior (CD); in the secondcase, we can think about a distributed routing and a distributed signaling

behavior (DD).

Now, suppose that the output of the PCE were: node 1 - link1 - node 2 - link 2 -

node 3 - link 3 - node 4 lambda 1. The ingress node prepares a GRSVP-TE

message, that is called Path message, and sends it to next node in the list; in

the example, node 1 sends a path message to node 2. The Node 2 checks

the lambda 1 on the link 2: if its still available, it forwards the request to the

next node in the list, that is, the node 3; otherwise, it sends a path error

message to the node 1 to inform that is not possible to satisfy the request. If

all is ok, the node 3 performs the same operations as the node 2 and forwardthe path message to the node 4.

The node 4 recognizes that its the egress node of the LSP, and, if all its ok,

builds a new message, the reservation message (RSV message) to send

backward to the ingress node, the node 1.

When the reservation message reaches the node 1, the LSP is created and the

information is sent to the NMS to align it about the status of the circuit.

Optionally, the ingress node can send a confirm message to the egress node to

inform it that all the LSP creation operation has been completed successfully.


40/56

40

Other GMPLS related concepts

Other GMPLS concepts:

Link Component (LC)

Link Cluster (LK)

Control Channel (CC)

Related to the GMPLS technology, there are some other important concepts:

the concepts of link component or LC;

the concepts of link cluster or LK;

the concept of control channel or CC.


41/56

41

Link Component

Link Component (LC)

What is a Link Component?

In GMPLS terminology a link component represents, in the control plane, a

physical link.A GMPLS Control Plane can use a physical bi-directional link if and only if it has

been declared as a link component; otherwise, for the WSON control plane, that

link doesnt exist.

In the picture, two WSON nodes are interconnected by a couple of fiber; from a

WSON point of view, they are interconnected by a single link component.


42/56

42

Link Cluster

LC 1

LC 2

Link Cluster (LK)

What is a Link Cluster?

ITU-T and IETF were worried about scalability problems that could have risen

due to the amount of links that a generic transport network could have.If there are many links in a WSON, more GOSPF-TE LSU messages are

required;

More LSU messages mean more load on the node controllers and more time for

PCEs to find out and validate best paths.

The idea has been to reduce the number of advertised links, by grouping a

number of physical links together, that is, a number of link components, in a

logical entity called link cluster; a link cluster is a cluster of link components.


43/56

43

Link Cluster

LC 1

LC 2

Link Cluster (LK)

Two link components can be inserted in the same link cluster only if they share

some common characteristics: same capacity, same set of SRLGs, colors,

administrative cost; same nodes as termination points; same kind of link levelprotection.

Its always necessary to create the logical entity link cluster, even if only one link

component is defined, because, the GOSPF-TE exchange information about link

clusters and not about link component.

During the control plane configuration phase, the normal procedure is:

Step 1: Define an empty link cluster, assigning to it all the GOSPF-TE

parameters, like administrative cost, SRLGs, colors, etc.

Step 2. Define one or more link components;

Step 3: Associate one or more link component to the already created link cluster.When all the link clusters are defined, even the one containing only one link

component, the GOSPF-TE process will only take them into consideration.

This means that, in general, the nodes all around the network are aware only of

the link clusters that are present in the network; The details about what is inside a

single link cluster are known only by the nodes that are the termination point of

that particular link cluster, through a specific GMPLS protocol called Link

Management Protocol or LMP.


44/56

44

the lmp protocol

Control Channel (CC) 2

Control Channel (CC) 1LMP Messages

LMP Messages

The Link Management Protocol of LMP has two major functionalities:

Setup and maintenance of the control channels defined between two nodes;

Setup and maintenance of the link cluster defined between two nodes.

Normally more than one control channel is defined between two nodes for

redundancy and load sharing; the LMP is responsible for managing these control

channels, for example, activating one secondary control channel, when the

primary is down.

The other important task of the LMP is related to link clusters: the correct

associations between link components and link clusters is maintained and

checked by the LMP. For example, if we try to associate two link components that

dont have common characteristics in a common link cluster, the LMP running

between the two nodes will forbid the operation.The LMP protocol is the glue that keeps control channel, link component and link

cluster together.


45/56

45

Protections

Lets now analyze the new advanced WSON protections in details.


46/56

46

WSON Protections

There are two different kinds of WSON protections:

The Transponder Sharing Protection or Green Protection

The Safe Optical Sub-Network Connection Protection (Safe OSNCP)

There are two different kinds of WSON protections:

The Transponder Sharing Protection or Green Protection;

The Safe Optical Sub-Network Connection Protection (Safe OSNCP).


47/56

47

transponder sharing

In case of Transponder Sharing Protection, the PCE calculates and validates

two or more LSPs:

One worker LSP, that is immediately signaled and activated by the ingress node;

One or more protection LSPs, that the ingress node do not signal and do not activate

immediately; they are maintained inside the database of the ingress node, associated to the

specific transponder sharing protected circuit.

The first kind of WSON protection is the transponder sharing protection: one

worker LSP is activated and one or more protection LSPs are pre-calculated and

pre-validated but not activated until a failure occurs.Only one transponder is necessary for this kind of protection;

This type of protection in similar to an On the Fly protection, in which, as soon as

a failure occurs, a new protection path is calculated, validated, signaled and

activated: the only difference is that, for the transponder sharing protection, the

paths calculation and validation is done in advance by the PCE off-line; only the

signaling and activation phase are done on the fly, when a failure occurs.

This is a flexible protection, because the protection bandwidth can be shared

among the protection paths of other WSON circuits.

The only disadvantage is the time for switching, not only because its necessaryto signal the new protection path, and this requires an amount of time that

depends on the number of nodes, but mainly because adding a new circuit in a

WDM network requires an amount of time for equalization and PMD

compensation. PMD compensation can be critical especially for high bit rates, like

40 Gigabit per seconds or 100 Gigabit per seconds.

Please notice that, in general, planning a WSON only with transponder sharing

protections, allows to save a lot of resources in terms of power consumption: this

is the reason why this type of protection is also called Green Protection.


48/56

48

transponder sharing

NODE 1 NODE 2 NODE 3

NODE 4NODE 5NODE 6

Link 1 Link 2

Link 3

Link 4

Link 5 Link6

Link 7

Link 8

Link9 Worker

LSP

Second

Protection

LSP

First

Protection

LSP

Third

Protection

LSP

Lets consider the transponder sharing example in the picture:

Suppose an operator would like to have a circuit from node 1 to node 4 that is

protected with a transponder sharing protection: one decision that he has to takeis how many protection path he wants the PCE to calculate and validate; lets, for

this case, suppose that the operator decided to have one worker LSP and three

protections LSPs.

Suppose also that we are in the case where the PCE runs off-line.

The operator passes his request to the PCE and the output of the PCE is a list of

four LSPs, one worker and three protections, all validated.

Now suppose that the PCE has decided the following paths:

For the worker: node 1- link 1- node 2- link 2- node 3- link 3- node 4 and the

lambda chosen is the green one;For the first protection: node 1- link 9- node 5- link 4- node 4 and the lambda

chosen is the red one;

For the second protection: node 1- link 8- node 6- link 7- node 5- link 4- node 4

and the lambda chosen is the yellow one;

For the third protection: node 1- link 1- node 2- link 5- node 5 link 4 node 4

and the lambda chosen is the purple one.

Please notice that the protection bandwidth is not reserved until a failure happen;

this means that the protection bandwidth can be shared with other protection

paths of other circuits.


49/56

49

transponder sharing


NODE 4NODE 5NODE 6

Link 1 Link 2

Link 3

Link 4

Link 5 Link

6

Link 7

Link 8

Link9

Now, suppose a failure happens on the link 2; the ingress node, the node 1 in this

case, checks if all the links on the path for the first protection LSP are ok; if all are

ok, the LSP is signaled and activated; if one of the links is not ok, the ingressnode checks the second LSP and, if all the links are ok, this LSP is signaled and

activated; over wise, the ingress node goes on checking all the possible

protection LSPs until one of these has all the links that are ok. In this case, the

first protection LSP is good because all its links are ok.

If, after a while, before repairing the link 2, a second failure occurs, on link 9, for

example, the third LSP is signaled and activated.

If there is an other failure, on link 8, for example, the ingress node is able to react

because it knows also a third protection LSP and the traffic can be protected.

Of course, in case of a fourth failure, the traffic is lost.

This type of protection is called transponder sharing because to implement this

protection only one transponder is necessary; if we would like to have the

capability not only to reroute the circuit, but also to change the lambda, it will be

necessary to use a kind of transponder that has the capability to tune its

frequency. Its common to refer to these transponders as tunable transponders.


50/56

50

transponder sharing

Line 1

Line 2

Line 9

Multi Direction Directionless Colorless

Line 1 Trp sharing

Protection

Now, lets see an example of a possible internal structure of a WSON node

(ROADM).

The configuration in the picture shows one card for each line direction; this cardis reconfigurable via software and is able to block or to allow to pass a particular

lambda; we refer to this type of card as Wavelength Selective Switch, or WSS,

card. The WSS card must be supported by a real-time equalizer, that can also be

embedded in the WSS card, depending on the particular implementation.

In a particular configuration that is called, directionless configuration, a WSS card

can be used to re-route a traffic coming from transponder.

In the example, we can see a circuit, with a transponder sharing protected circuit

using a tunable transponder, that is able to change the frequency to the WDM

link via software configuration, guided by the node controller.

In the starting situation the circuit uses the orange lambda and passes through

the line 1.

Now, lets concentrate on what happens in case of a fault.


51/56

51

transponder sharing

Line 1

Line 2

Line 9

Multi Direction Directionless Colorless

Line 1 Trp sharing

Protection

A failure occurs on the line 1: the controller immediately activates its GRSVP-TE

process to signal and to activate the pre-calculated protection LSP passing

through the line 9.Please notice that, in this example, we changed the lambda, because the orange

lambda on the line 9 has been considered as already used by an other circuit; the

change of lambda is obtained using a tunable transponder, in a configuration

called color-less, that means, with no fixed lambda.

If, at this point, a second failure occurs on the line 9, a second pre-calculated

LSP is signaled and activated.


52/56

52

Safe OSNCP

In case of the Safe OSNCP Protection, the PCE calculates and validates three

LSPs:

One worker LSP and one protection LSP, that are both immediately signaled and activated

by the ingress node;

One more protection LSP, that the ingress node doesnt signal and doesnt activate

immediately; its maintained inside the database of the ingress node, associated to the

specific Safe OSNCP protected circuit.

The second kind of WSON protection is the Safe OSNCP protection: one worker

LSP and one protection LSP are activated and one protection LSP is pre-

calculated and pre-validated but not activated until a failure occurs.Two transponders are necessary for this kind of protection;

The traffic is sent in both the worker and protection side at the same time, as it

would happen for the traditional OSNCP protection.

In case there is a failure on the worker side, the traffic is switched on the

protection side in less than 50 milliseconds and the second protection LSP is

immediately signaled and activated.

Now, in case there is a second failure, this time on the first, original, protection

side the traffic is switched on the second protection LSP in less than 50

milliseconds.


53/56

53

SAFE OSNCP


NODE 4NODE 5NODE 6

Link 1 Link 2

Link 3

Link 4

Link 5 Link

6

Link 7

Link 8

Link9 Worker

LSP

First

Protection

LSP

Second

Protection

LSP

Lets consider the example in the picture:

Suppose an operator would like to have a circuit from node 1 to node 4 that is

protected with a Safe OSNCP protection.Also in this case, suppose that we are in the case where the PCE runs off-line.

The operator passes his request to the PCE and the output of the PCE is a list of

three LSPs, one worker and two protections, all validated.

Now suppose that the PCE has decided the following paths:

For the worker LSP: node 1- link 1- node 2- link 2- node 3- link 3- node 4 and the

lambda chosen is the green one;

For the first protection LSP: node 1- link 8- node 6- link 7- node 5- link 4- node 4

and the lambda chosen is the orange one;

For the second protection LSP: node 1- link 9- node 5- link 4- node 4 and thelambda chosen is the red one.


54/56

54

Safe OSNCP

First

Protection

Side

Worker

Side

The traffic switch on this side in less than 50 ms and the pre-calculated secondary protection LSP is signaled and activated


NODE 4NODE 5NODE 6

Link 1 Link 2

Link 3

Link 4

Link 5 Link

6

Link 7

Link 8

Link9The traffic is switched on this side in less than 50 ms

The traffic is LOST

The output of the PCE is passed through the NMS to the ingress node of the

circuit, in this example, the node 1, but only the worker LSP (green) and the first

protection LSP (orange) are signaled and activated; the other protection LSP ismaintained in the database of the node associated to this particular circuit.

Now suppose that a failure happens on the link 2: in less than 50 milliseconds the

traffic is switched on the first protection side (orange); in the meanwhile, the

secondary protection LSP (red) is signaled and activated.

If, after a while, and before repairing the link 2, a second failure happens on link

7: the ingress node reacts switching the traffic on the second protection side in

less than 50 milliseconds.

If a third failure occurs, for example, on the link 9, the traffic is lost until one of the

LSPs is repaired.


55/56

55

Safe OSNCP

Safe-1+1OSNCP/OUPSR

Multi Direction Directionless ColorlessProtection

Line 1Line 1

Line 2

Line 9

Lets see an example of what happens inside a WSON node in case of a SAFE

OSNCP protection.

In the picture, its possible to see one circuit formed by two LSPs: the orange linerepresents the worker LSP and the green line represents the protection LSP; two

tunable transponders are necessary for the SAFE OSNCP protection.

Suppose that a failure occurs on the line 1: The traffic is switched on the

protection side in less than 50 milliseconds, as it would happen for the normal

OSNCP protection. The original worker path is deleted; a new pre-calculated

protection LSP 2 is signaled and activated passing through the line 2.

Now suppose that an other failure happens on line 9. The traffic is switched on

the LSP 2 in less than 50 ms.

In case a third failure occurs, the circuit is not protected anymore, because onlytwo protection LSPs are pre-calculated by PCE for that specific circuit. In

conclusion, the Safe OSNCP protection is very fast compared with the

Transponder sharing protection; we pay this speed with more bandwidth to

reserve for a specific circuit, because the protection bandwidth cannot be shared

with other protection LSPs of other circuits and with an additional transponder

that must be present.


56/56

Thank you for taking the time to listen to this ASON-WSON Fundamentals

course.

For more technical information regarding WDM and OTN, please view therelevant Fundamentals courses.

ason-wson fundamentals rev9

Documents