AN HIGH PERFORMANCE TRAFFIC ENGINEERING MODEL

FOR ALL-OPTICAL NETWORKS

Evolutionary GMPLS control plane services in all-optical cross-connects

Francesco Palmieri

Centro Servizi Didatico Scientifico, Federico II University, Via Cinthia 45, Napoli, Italy

Keywords: Internet services, Dial-up networking

Abstract: One of the major issues in the networking industr

y today is the tremendous demand for more and more

bandwidth. With the development of all-optical networks and the use of Dense Wavelength Division Multi-

plexing (DWDM) technology, a new and probably very crucial milestone is being reached in network evolu-

tion. In this scenario, carriers need powerful, commercially viable and scalable tools that can be used to bal-

ance the traffic load on the various fiber links/wavelengths and optical switches in the network so that none

of these components is over utilized or underutilized. Generalized Multi-Protocol Label Switching (GMPLS)

is actually the most promising technology, which will play a key role in future IP pure optical networks by

providing the necessary bridges between the IP and optical layers to deliver effective traffic engineering

features and allow for interoperable and scalable parallel growth in the IP and photonic dimension. This pa-

per propose an integrated control plane approach that will combine existing GMPLS control plane tech-

niques with the point-and-click provisioning capabilities of photonic switches to set up optical channel trails

and to distribute optical transport network topology state information. The GMPLS control plane will sup-

port various traffic engineering functions, and enable a variety of protection and restoration capabilities,

while simplifying the integration of photonic switches and label switching routers.

1 INTRODUCTION

With the development of an information-oriented

society, the explosive growth of the Internet and the

emerging demand for integrated network-based ap-

plications, the needs of network capacity have in-

creased dramatically, requiring a substantially higher

bandwidth than that offered by current networks

based on Electronic Time Division Multiplexing

(ETDM) technology. For this reason, all-optical

networks, based on the concepts of Wavelength Di-

vision Multiplexing (WDM) and wavelength rout-

ing, promising data transmission rates several orders

of magnitude higher than current networks, are con-

sidered as the transport networks for the future

(Mukherjee, B., 1997). In such networks, two adja-

cent nodes are connected by one or multiple fibers

each carrying multiple wavelengths/channels. Each

node consists of a dynamically configurable optical

cross-connect (OXC) which supports fiber switching

and wavelength switching, that is, the data on a

specified input fiber and wavelength can be switched

to a specified output fiber on the same wavelength

(Mokhtarm A. et al., 1998). In order to transfer data

between source–destination node pairs, a lightpath

needs to be established by allocating the same wave-

length throughout the route of the transmitted data.

Benefiting from the development of all-optical am-

plifiers, lightpaths may span more than one fiber link

and remain entirely optical from end to end. It has

been demonstrated that the introduction of wave-

length-routed networks not only offers the advan-

tages of higher transmission capacity and better

switching throughput, but also satisfies the growing

demand for protocol transparency and simplified

operation and management (Karasan, E., 1998). In

this scenario, OXCs are likely to emerge as the pre-

ferred option for switching multi-gigabit or even

terabit data streams, since the slow electronic per-

packet processing is avoided. The most interesting

and desirable function of an OXC is to dynamically

reconfigure the network at the wavelength level for

restoration or to dynamically accommodate changes

in bandwidth demand. OXC systems are expected to

be the cornerstone of the photonic layer providing

carriers more dynamic and flexible options in build-

ing network topologies with enhanced survivability.

311

Palmieri F. (2004).

AN HIGH PERFORMANCE TRAFFIC ENGINEERING MODEL FOR ALL-OPTICAL NETWORKS - Evolutionary GMPLS control plane services in

all-optical cross-connects.

In Proceedings of the First International Conference on E-Business and Telecommunication Networks, pages 311-316

DOI: 10.5220/0001397103110316

 SciTePress

The process of adaptively mapping traffic flows onto

the physical topology of a network and allocating

resources to these flows, usually referred to as traffic

engineering, is one of the most difficult tasks facing

service providers today. In this way, they can exploit

some economies on the bandwidth that has been

provisioned across the entire network and increase

their revenues by fully supporting the needs of time

or/and mission critical applications. Traffic engi-

neering (TE) should be viewed as assistance to the

routing/switching infrastructure that provides addi-

tional information used in forwarding traffic along

alternate paths across the network, trying to optimize

service delivery throughout the network by improv-

ing the network utilization and avoiding congestion

caused by uneven traffic distribution. Traditionally,

all provisioning and engineering in optical networks

has required manual planning and configuration,

resulting in setup times of days or even weeks and a

marked reluctance amongst network managers to de-

provision resources, when it implies impacts on

other services. Where control and traffic manage-

ment protocols have been deployed to provision op-

tical networks they have been proprietary and have

suffered from interoperability problems. Recently, a

new paradigm for the design of control planes for

OXCs intended for automatically switched optical

transport networks was proposed in (Banerjee, A. et

al., 2001). This new paradigm is termed GMPLS and

exploits recent advances in MPLS control plane

technology to foster the expedited development and

deployment of a new class of versatile OXCs that

specifically address the optical transport needs of the

Internet. The GMPLS control plane has been shown

to be an extensible general purpose control plane

technology, supporting flexible traffic engineering,

for a variety of high performance network elements.

This paper propose an integrated GMPLS-based

framework built on the strengths of MPLS for fine

grain traffic load balancing and optical layer re-

configuration for providing effective TE in all-

optical networks. It is modeled according to the

principle of providing a single control plane, directly

suitable for OXC, for the transport and service man-

agement layers and using widely available IP traffic

engineering and management tools to greatly sim-

plify and scale network operations. Beyond eliminat-

ing proprietary vendor “islands of deployment”, this

common control plane, based on a single set of se-

mantics, enables independent innovation curves

within each product class and faster service deploy-

ment with end-to end provisioning in the whole op-

tical network.

2 THE GMPLS PARADIGM

GMPLS has been proposed shortly after MPLS.

With the success of MPLS in TE and resource man-

agement of packet switched IP networks, optical

network providers have driven a process to general-

ize the applicability of MPLS to cover all-optical

networks as well. In GMPLS the idea of a label can

be generalized to be anything that is sufficient to

identify a traffic flow. GMPLS mainly focuses on

the control plane that performs connection manage-

ment for the data plane, by creating label switched

paths (LSPs) on both packet-switched and non-

packet-switched (i.e. lambda-switched) interfaces. In

detail, it realizes four basic functions as follows:

− Routing control: It provides the routing ca-

pability, TE and topology discovery.

− Resource discovery: It provides a mechanism

to keep track of the system resource avail-

ability such as bandwidth, multiplexing capabil-

ity, and ports.

− Connection management: It provides end-to-end

service provisioning for different services. This

includes connection creation, modification,

query, and deletion.

− Connection restoration: It provides an addi-

tional level of protection to the networks.

GMPLS encompasses control plane signaling for

multiple interface types: Packet Switch Capable

(PSC), Time Division Multiplexing Capable (TDM),

Lambda Switch Capable (LSC) and Fiber Switch

Capable (FSC). Each of them can be considered as a

lower level LSP endpoint nested within a higher-

level LSP one. In this sense the extension of MPLS

for supporting different types of LSP is the generali-

zation of the stacking functionality. The diversity of

controlling not only switched packets and cells, but

also TDM network traffic and optical network com-

ponents makes GMPLS flexible enough to position

itself in the direct migration path from electronic to

all-optical network switching. In order for these

interface types and link bundles to be handled ac-

cordingly, GMPLS needed a method to manage the

links between adjacent nodes. The Link Manage-

ment Protocol (LMP) was developed to address sev-

eral link specific problems that surfaced when gen-

eralizing the MPLS protocol across different inter-

face types. LMP provides control channel manage-

ment, link connectivity verification, link property

correlation, and fault isolation. Control channel

management establishes and maintains connectivity

between adjacent nodes using a keep-alive protocol.

Link verification verifies the physical connectivity

between nodes, thereby detecting loss of connections

ICETE 2004 - SECURITY AND RELIABILITY IN INFORMATION SYSTEMS AND NETWORKS

312

and misrouting of cable connections. Fault isolation

pinpoints failures in both electronic and optical links

without regard to the data format traversing the link.

3 TE IN PHOTONIC NETWORKS

If a TE application implements the right set of fea-

tures, it should provide precise control over the

placement of traffic flows within a routing and

switching domain gaining better network utilization

and realizing a more manageable network.

3.1 Fundamental TE requirements

A traffic engineering solution suitable for pure opti-

cal networks will always consist of a number of ba-

sic functional components, as described below:

− Traffic monitoring, analysis and aggregation: is

responsible for collecting traffic statistics from

the network elements, e.g. the OXCs. Then the

statistics are analyzed and/or aggregated to pre-

pare for the traffic engineering and network re-

configuration related decision-making.

− Bandwidth demand projection: projects the

bandwidth requirements in the near future based

on past and present measurements and the char-

acteristics of the traffic arrival processes. The

bandwidth projections are used for subsequent

bandwidth allocation.

− Reconfiguration trigger: consists of a set of

policies that decide when a network level recon-

figuration is performed. This is based on traffic

measurements, bandwidth predictions, and on

operational issues, e.g., to suppress influence of

transitional factors and reserve adequate time

for network to converge.

− Topology design: provides a network topology

based on the traffic measurements and predic-

tions. Conceptually this can be considered as

optimizing a graph (i.e. OXC connected by light

paths in the WDM layer) for specific objectives

(e.g. maximizing throughput), subject to certain

constraints (e.g. nodal degree, interface capac-

ity), for a given load matrix (i.e. traffic load ap-

plied to the network.) This is in general a NP-

hard problem. Since reconfiguration is regularly

triggered by continually changing traffic pat-

terns, an optimized solution may not be stable.

It may be more practical to develop heuristics

that emphasize more on factors like fast conver-

gence, and less impacts on ongoing traffic, than

on optimality.

− Topology migration: consists of algorithms to

coordinate the network migration from an old

topology to a new topology. As WDM recon-

figuration deals with large-capacity channels,

changing allocation of channel resources in this

coarse granularity has significant effects on a

large number of end user flows. Traffic flows

have to adapt to the light path changes at and af-

ter each migration step. These effects can poten-

tially spread over the routing pattern of the net-

work, which in turn may affect more user flows.

3.2 TE in current overlay networks

Today’s carrier-class data services network typically

consists of five layers (Fig. 1): a local access layer

connecting enterprise networks to the carrier net-

work, IP for carrying data applications, ATM for

traffic engineering, SDH/SONET for transport, and

wavelength division multiplexing (WDM) for capac-

ity. IP supports the wide variety of data and multi-

media applications dominating traffic growth in the

networks. ATM supplies the QoS and service guar-

antees for reliability and provides the means for en-

gineering the flows of traffic. SDH/SONET net-

works, grooming optical channels onto light paths,

are pervasive in their ability to distribute traffic re-

liably within the Metropolitan (MAN) and Wide

Area Network (WAN). Carriers implement WDM

extensively to augment the capacity of their installed

base of fiber by combining multiple optical paths

onto a single fiber optic link.

Figure 1: Multi-layered Network Architecture

Carriers manage each layer separately and use

each layer to perform one function as well as possi-

ble. Each layer relies on a different control and data

plane technology and has its own network manage-

ment system. The SDH/SONET and WDM layers

are static network layers that must be manually pro-

visioned. The IP layer gives the carrier some auto-

mation in traffic distribution and service provision-

ing but provides little in the TE control or QoS guar-

antees. Consequently, carriers must use manual

provisioning to implement TE policies and to ad-

dress performance deficiencies. Thus, the layered

network design leads to frequent manual reconfigu-

AN HIGH PERFORMANCE TRAFFIC ENGINEERING MODEL FOR ALL-OPTICAL NETWORKS - Evolutionary

GMPLS control plane services in all-optical cross-connects

313

rations of the SDH/SONET and WDM layers for IP-

related services. The static nature of the current car-

rier network architecture left many carriers unpre-

pared for provisioning large amounts of fiber and

optical wavelengths to meet the needs of enterprise

IP networks, ISPs, VPNs, and other data-centric

services. Manual provisioning left the carriers with

a complex collage of multiplexers, OXCs, and fiber

patch panels. Customers want rapid provisioning

and flexible deployment of services, but the multi-

layer control structure of carrier networks built dur-

ing the last decade could not cope with the rapidly

growing and changing service demands. Carriers

need new methods for provisioning optical band-

widths and for automatically provisioning shared-

bandwidth across their networks. They need to sup-

port IP data traffic effectively, provide QoS guaran-

tees for customer service level agreements (SLAs),

and utilize core network bandwidth efficiently.

Much of the current carrier network architecture

lacks the ability to satisfy the needs of a growing

variety of IP applications. In order to bypass the

tradeoffs due to excessive layering in traffic control

the basic and necessary TE functions must move

directly to the OXCs and WDMs. At the end, this

results in a simpler, more cost-efficient network that

will transport a wide range of data streams and very

large volumes of traffic. Recently, an innovative TE

framework (Banerjee, A. et al., 2001) built on the

strengths of MPLS for fine grain traffic load balanc-

ing, and optical layer re-configuration has been pro-

posed. Moreover, there is an approach to the design

of control planes for optical cross-connects which

leverage existing control plane techniques developed

for MPLS TE. This approach combines recent ad-

vances in MPLS traffic engineering control plane

constructs with optical cross-connect technology to

provide a framework for real-time provisioning of

optical channels, foster development and deploy-

ment of a new class of optical cross-connects, and

allow the use of uniform semantics for network

management and operations control in networks

consisting of IP addressable and TE-capable optical

cross-connects.

4 INTEGRATING GMPLS

CONTROL PLANE IN OXCS

All of the above observations suggest, therefore, that

the GMPLS Traffic Engineering control plane would

be, with some minor extensions, very suitable as the

control plane for OXCs. This concept originated

from the observation that from the perspective of

control semantics, an OXC with an GMPLS Traffic

Engineering-enabled control plane would resemble a

Label Switching Router, subsuming and spanning

LSRs and OXCs functionalities in a single integrated

control plane, with some restriction due to the pecu-

liarity of the OXC data plane. In fact, the adaptation

of MPLS control plane ant TE concepts to OXCs,

which results in OXC-LSRs, needs to consider and

reflect the domain specific peculiarities of the OXC

data plane. From a data plane perspective, an LSR

switches packets according to the label that it car-

ries. An OXC uses a switching matrix to connect an

optical wavelength/signal from an input fiber to an

output fiber. From a control plane perspective, an

LSR bases its functionality on a table that maintains

relations between incoming label/port and outgoing

label/port. It must be pointed that in the case of the

OXC, the table that maintains the relations is not a

software entity but it is implemented in a more

straightforward way, e.g. by appropriately configur-

ing the micro-mirrors of an optical switching fabric.

There are several constraints in re-using the GMPLS

control plane. These constraints arise from the fact

that LSRs and OXCs use different data technologies.

More specifically, LSRs manipulate packets that

bear an explicit label and OXCs manipulate wave-

lengths that bear the label implicitly. That is, since

the analogue of a label in the OXC is a wavelength

or an optical channel there are no equivalent con-

cepts of label merging nor label push and pop opera-

tions in the optical domain. The transparency and

multi-protocol properties of the MPLS Control Plane

approach would allow an OXC to route optical

channel trails carrying various types of digital pay-

loads (including IP, ATM, SDH, etc) in a coherent

and uniform way. The distribution of topology state

information, establishment of optical channel trails,

all-optical network Traffic Engineering functions,

and protection and restoration capabilities would be

facilitated by the GMPLS control plane. An out-of-

band IP communications system can be used to carry

and distribute control traffic between the control

planes of different connected OXCs, perhaps

through dedicated supervisory channels, using dedi-

cated wavelengths or channels, or an independent

out-of-band IP network. An OXC that uses the

GMPLS control plane would effectively become an

IP addressable device. Thus, this proposition also

solves the problem of addressing for OXCs. In this

environment, SNMP, or some other network man-

agement technology, could be used for element

management. A reasonable architectural model for

an OXC equipped with an integrated GMPLS con-

trol plane (OXC-LSR) consists of two components:

the data forwarding plane and the GMPLS control

plane. A simple schema, consistent with IETF

GMPLS standard draft, is represented in Fig. 2 be-

low:

ICETE 2004 - SECURITY AND RELIABILITY IN INFORMATION SYSTEMS AND NETWORKS

314

Figure 2: OXC with GMPLS control plane architecture

4.1 The OXC-LSR data plane

The data/forwarding plane performs data routing to

the appropriate ports, channel add/drop to external

legacy networks (using the edge interfaces) and la-

bel/lambda swapping through an array of demulti-

plexers, wavelength converters, optical cross-

connects, optical amplifiers and multiplexers. A

simple and linear architectural model for the

data/forwarding plane of an all-optical OXC sup-

porting a GMPLS-based control plane is shown in

Fig. 3 below:

Figure 3: All-optical OCX data plane architecture

Here, the WDM Demultiplexers separate incoming

wavelengths (N grouped lambdas) from input ports

into individual lambdas. The wavelength converters

will perform, if necessary, wavelength conversion

(that is swapping the optical label in the GMPLS

control plane) based on the instruction from the con-

trol plane. A sufficiently large low-loss connectivity

and compact design all optical switching fabric can

be realized by using the reflection of light and mi-

cro-electromechanical systems (MEMS) technology,

now widely available on the market. This MEMS-

based multi-layer optical switching fabric driven by

a micro-machined electrical MEMS actuator redi-

rects, according to the GMPLS control plane instruc-

tions, each wavelength into appropriate output ports

passing through optical amplifiers, typically Erbium

doped Fiber Amplifiers (EDFA) or Silica Erbium

based Dual band Fiber (BDFA) amplifiers, which

boost the signal power inline without the need for

any optoelectronic conversion to cope with the ef-

fects of polarization mode dispersion and attenuation

on long distances. The WDM Multiplexer then

groups the wavelengths from the above multiple

layers of cross connects. There may be a special

input/output port that adds and drops from the edge

of the all-optical network acting as the “interface”

between the OXC-LSR and external legacy net-

works such as ATM, SONET/SDH and Gigabit

Ethernet. The channels that can be added/dropped

can be configured automatically. By definition, the

resulting Optical Cross Connect architecture is

strictly non-blocking and transparent to bit-rate and

data format, due to its all-optical implementation. It

is also link modular since the addition of new in-

put/output fibers just requires the addition of new

elements without changing the OXC overall struc-

ture. On the other hand the OXC architecture is not

wavelength modular, since the adding of a new

channel changes all the used MUXes, DEMUXes,

MEMS fabric and wavelength converters.

4.2 The OXC-LSR control plane

The OXC-LSR GMPLS control plane has to perform

the following functions in order to set up the G-LSP

routing and resource table:

− Wavelength Assignment and Routing Manage-

ment at each link.

− Resource management and Traffic Engineering

to set up the generalized LSPs (or G-LSPs)

− Link Management between OXC-LSRs (with

the LMP protocol)

Routing Management and wavelength assignment:

In order to transfer data between source–destination

node pairs, a lightpath needs to be established by

allocating the same wavelength throughout the route

of the transmitted data. As In general, if there are

multiple feasible wavelengths between a source

node and a destination node, then a wavelength as-

signment algorithm is required to select a wave-

length for a given lightpath. The wavelength selec-

tion may be performed either after a route has been

determined, or in parallel with finding a route. Since

the same wavelength must be used on all links in a

lightpath, it is important that wavelengths are chosen

in a way which attempts to reduce blocking for sub-

sequent connections. When lightpaths are estab-

lished and taken down dynamically, routing and

wavelength assignment decisions must be made as

connection requests arrive to the network. It is pos-

sible that, for a given connection request, there may

be insufficient network resources to set up a light-

path, in which case the connection request will be

blocked. The connection may also be blocked if

AN HIGH PERFORMANCE TRAFFIC ENGINEERING MODEL FOR ALL-OPTICAL NETWORKS - Evolutionary

GMPLS control plane services in all-optical cross-connects

315

there is no common wavelength available on all of

the links along the chosen route. Thus, the objective

in the dynamic situation is to choose a route and a

wavelength which maximizes the probability of set-

ting up a given connection, while at the same time

attempting to minimize the blocking for future con-

nections.

Resource management and Traffic Engineering:

In order to set up a lightpath, a signaling protocol is

required to exchange control information among

nodes, to distribute labels and to reserve resources

along the path. In our case, the signaling protocol is

closely integrated with the routing and wavelength

assignment protocols. Suitable GMPLS signaling

protocols for our model include RSVP and CR-LDP.

Each of them can be used to instantiate the optical

channel trails. With the RSVP extensions, for exam-

ple, the wavelength information or optical channel

information implicitly referring to the label concept,

will be used to control and reconfigure the OXCs.

Furthermore, each node maintains a representation

of the state of each link in the network. The link

state includes the total number of active channels,

the number of allocated channels, and the number of

channels reserved for lightpath restoration. Addi-

tional parameters may be associated with allocated

channels, for example, some lightpaths may be pre-

emptable or have associated hold priorities. Once the

local inventory is constructed, the node engages in a

routing protocol to distribute and maintain the topol-

ogy and resource information. Standard IP routing

protocols, such as Open Shortest Path Forwarding

(OSPF) or Intermediate System-Intermediate System

(IS-IS) with GMPLS TE extensions, may be used to

reliably propagate the information. Furthermore, the

OXC will maintain a WFIB (Wavelength Forward-

ing Information Base) per interface (or per fiber).

This is because lambdas and/or channels (labels) are

specific to a particular interface (fiber), and the same

lambda and/or channel (label) could be used concur-

rently on multiple interfaces (fibers).

Link Management between OXC-LSR: Although

LMP assumes the messages are IP encoded, it does

not dictate the actual transport mechanism used for

the control channel. However, the control channel

must terminate on the same two nodes that the

bearer channels span. As such, this protocol can be

implemented on any OXC, regardless of the internal

switching fabric. A requirement for LMP is that each

link has an associated bi-directional control channel

and that free bearer channels must be opaque (i.e.,

able to be terminated); however, once a bearer chan-

nel is allocated, it may become transparent. Note

that this requirement is trivial for optical cross-

connects with electronic switching planes, but is an

added restriction for photonic switches.

5 CONCLUSIONS

In this paper the key aspects related to Traffic Engi-

neering in new generation all-optical network infra-

structure are reviewed investigating how the MPLS

Traffic Engineering-enabled control plane could be

adapted and reused as the control plane for optical

cross-connects. Such a control plane would be used

to distribute optical transport network topology state

information and to setup optical channel trails. Such

a control plane would support various traffic engi-

neering functions in the optical domain, and enable a

variety of protection and restoration capabilities.

Furthermore, such a control plane technology would

expedite the development and deployment of a new

class of versatile IP-addressable OXCs. We envision

a horizontal network where all network elements

work as peers to dynamically establish optical paths

through the network. This new photonic internet-

work will make it possible to provision high band-

width in tenths of seconds, enable new revenue-

generating services, and dramatic cost savings for

the service provider.

REFERENCES

Mukherjee, B., 1997. Optical Communication Networks.

Mc-Graw-Hill.

Karasan, E., Banerjee, A., 1998. Performance of WDM

transport networks. In IEEE Journal Seleced Areas in

Communications, vol. 16, pp. 1081–1096.

Banerjee, A., Drake, J., Lang, J., Turner, B., Awduche, D.,

Berger, L., Kompella, K. and Rekhter, Y., 2001. Gen-

eralized Multiprotocol Label Switching: An Overview

of Signaling Enhancements and Recovery Techniques.

In IEEE Communications Magazine.

Banerjee, A., Drake, J., Lang, J., Turner, B., Awduche, D.,

Berger, L., Kompella, K. and Rekhter, Y., 2001. Gen-

eralized Multiprotocol Label Switching: An Overview

of Routing and Management Enhancements. In IEEE

Communications Magazine.

Mokhtarm A. and Azizoglu, M., 1998. Adaptive wave-

length routing in all-optical networks. In IEEE/ACM

Transactions on Networking, vol. 6, pp. 197–206.

ICETE 2004 - SECURITY AND RELIABILITY IN INFORMATION SYSTEMS AND NETWORKS

316