ViewsLetter(SM) on Provisioning

Site Map



Flanagan Consulting                 Network Analysts and Consultants
                                         "We Have the Experience"
ViewsLetter on Provisioning                10 Feb 2003        #16


By V. Kaminsky, PracTel, Inc.
Associate, Flanagan Consulting

Despite setbacks in the development of fiber optic network infrastructure,
a number of vendors still continue to work on the optical network
“automation” process and the development of intelligent networks.  Optical
switches from these vendors are intelligent enough to at least:
·    Learn the topology of a set of nodes, automatically;
·    Set up mesh-type network protection;
·    Automate provisioning (based on signaling);
·    Deal with digital “lambdas” (separate wavelengths that each carry 2.5
Gbit/s, 10 Gbit/s, or 40 Gbit/s);
·    Groom lambdas from tributary links;
·    Work with Dense Wave Division Multiplexing (DWDM), allowing signal
transmission of thousands kilometers without regenerating the signal

Multiple studies of optical switch technologies support the need for these
features and provide many reasons why these technologies have value, why
they have to be developed and commercialized--even in times when demand
for these switches is almost non-existent.  Some references can be found
in ViewsLetter #5 “Lambda on Demand in Optical Core may Lead Automation of
Carrier Nets.”

This paper provides an overview of key products for intelligent optical
networks, the core optical switches.  The scope was limited by a number of
factors.  The most serious problem is that, at present, optical
intelligent networks are still mostly in development and standardization
stages.  The client base for optical switches is very small.  Rather than
provide details, this paper intends to serve as a guide for those working
or interested in this area.

Note that ITU recommendations G.872 and G.709 use the term “Optical
Transport Networks.”  In this paper, term “intelligent” stresses the
intent of optical network developers to make these networks
“self-learning” and “self-organized”;  i.e., to enable them to react
effectively to changes in customer demand as well on changes in network
topology or connectivity.

Initially, the main purpose of optical switches was to support SONET/SDH
ring-based architectures, and to allow interconnection of multiple rings.
One of the first attempts to create a platform to interconnect multiple
optical rings has been made by Lucent (Bandwidth Manager, BWM).  It was
commercially introduced in the late 1990s.  This platform was deployed in
a number of networks, mostly supporting undersea cables.  It has a very
large footprint, and lost its market share to a more compact product from
Ciena, the CoreDirector.  Its smaller footprint brought further success
for optical switching.

BWM also created problems with ring interconnection when a network
required support of traffic between stacked rings (that is, multiple rings
on the same path but not initially interconnected).  More recently, there
were the first reports of successful mesh optical networks in which these
switches provide intelligence for self-healing, self-organization, and
automatic provisioning.

In general, the market for intelligent optical networks, supported by
ITU-T standards (G.709, Automatic Switched Optical Network (ASON), G.872,
and others), continues to expand, although at a much slower pace than was
predicted initially.  The bases for such networks are various flavors of
optical switching platforms that, in general:
·    Interconnect lambdas for 2.5 Gbit/s, 10 Gbit/s and 40 Gbit/s signals;
·    Groom tributaries for these lambdas;
·    Provide at least STS-1 or STM1 switching granularity;
·    Interconnect multiple rings: Multiplex Section Shared Protection Ring
(MS-Spring),  Bi-directional Line Switched Ring (BLSR), Unidirectional
Path Switched Ring (UPSR);
·    Protect tributaries using various schemes;
·    In later versions, allow mesh-topology protection;
·    For mesh topologies, discover spare capacity automatically, and optimize
its use;
·    Implement various self-healing algorithms;
·    Automate and optimize provisioning process;
·    Configure either SONET or SDH in software;
·    Connect directly with DWDM systems.

This paper divides optical switches into two major groups:  one with
optical-electrical-optical (OEO) conversion, and another with pure optical
processing of signals.  Note that recently a hybrid solution has been
developed with a platform containing two switching fabrics, one OEO and
one pure optical.

These switches convert optical signals to electrical signals at the switch
input, switch electrically, and convert electrical signals back to optical
at the output of the device.

This company commercially introduced the OPTera Connect HDX Optical
Switch.  The initial switch matrix size of 320 Gbit/s or 640 Gbit/s can
expand in a future model to 3.84 Tbit/s.  It has a compact design, and
with initial switching capacity one shelf of the standard NEBS- or
ETSI-compliant bay can host the whole switch.  It has granularity of
STS-1, and processes lambda sizes of 10 Gbit/s (40 Gbit/s in later
versions).  On the drop side, the switch handles a number of SONET and SDH
signal speeds, starting from OC-3/STM-1.  Among other features, it allows
direct DWDM interfaces on the line side.  Nortel is also advertising a
smaller version of HDX, called DX.

Nortel announced its "Advanced Network Intelligence" will deliver, in the
future, full topology auto-discovery, together with pre-planned mesh
restoration.  It also allows dynamic connection provisioning.  This
solution features an optical control plane, providing customized,
adaptable optical services to bandwidth users.  The solution is based on
the ASON standard architecture and the routing and signaling protocols
associated with Generalized Multiprotocl Label Switching (GMPLS).  Special
planning tools will allow users to simulate, plan, and optimize mesh

This company stopped developing the OOO switch.

The LamdaUnite MultiService Switch (MSS) was introduced to the market a
couple of years ago.  It has many features similar to the Nortel product.
Initially a 160 Gbit/s switching matrix, it can be upgraded while in
service to the standard configuration of 320 Gbit/s.  A doubling of this
capacity is expected in the near future.  Its distinguishing features
(among others) are implementation of a Trans Oceanic Protocol (TOP) in a
10 Gbit/s 4F MS-Spring configuration (reducing switching time for very
long undersea networks) and introduction of the DS-3 interface.  MSS can
comprise one or more Terminal Multiplexers, or ADM functions, in a single
node.  It also can act as a fully non-blocking cross connect.

Lucent is planning to use a control plane called Synchronous Network
Navigator (SNN) for mesh networks.  As a part of the intelligent network
platform, SNN will automate SONET/SDH connection set-up, provide fast
restoration, and support automatic discovery of the network topology.  SNN
will discover available transport capacity and routes, and provide
connection management.  MSS will support the signaling necessary to
actively participate in the SNN.

This company commercially introduced an OOO platform, LambdaRouter, but
discontinued it quickly.

This company has different lines of products for SONET and SDH
technologies.  The Alcatel Core Node Solution consists of an optimized
combination of DWDM systems, lambda management, SONET/SDH multiservice
nodes and gateways (Alcatel's term for optical switches), and a control
plane based on a GMPLS protocol and ASON standards (in the future).  It
has a centralized algorithm for automatic discovery of spare network
capacity.  By taking into account a priority level assigned to each
network connection and the status of other network resources, a
path-selection algorithm expertly regulates access to the shared capacity
and restores any path affected by a failure without operator intervention.
 This plane also will support automatic provisioning by signaling.

Alcatel 1674 Lambda Gate does not have a 40 Gbit/s interface.  In R1.0, it
occupies two racks with limited functionality of 2.5 Gbit/s and 10 Gbit/s
interfaces on both line and drop sides.  A typical application
interconnects three 10 Gbit/s rings from which 20 Gbit/s ports can be
dropped.  In R1.1, the switching fabric can go up to 5 Tbit/s, and offers
the additional function of the ITU-T G.709 OCh (Optical Channel)
interface.  The control plane in this release will support a mesh network
topology as well as rings.

The 1677 Sonet Link supports SONET applications.

Sycamore's SN 16000 switch is a software-centric, highly integrated
switching platform that brings advanced capabilities and manageability to
intelligent optical networking, from the metro edge to the optical core.
Sycamore is one of the companies that made progress in GMPLS-based control
plane development and implementation.  This company has demonstrated
interoperability of SN 16000 with Cisco, Juniper, and other IP routers in
a mesh topology.

The switch provides capacity up to 160 Gbit/s in a single shelf of the
standard bay.  Its management software supports maximum operational
efficiency and multi-dimensional scalability.  Network-aware intelligence
enables end-to-end provisioning and a business-driven migration toward
optical mesh architecture, without compromising the reliability of
existing services.  Fully integrated, Sycamore's software-rich and
scalable suite of BroadLeaf and SILVX SW form a foundation for a flexible,
dynamic optical infrastructure - keeping a network open to new

The SN 16000 switch is ready to adopt evolving standards from ITU and
others for a flexible control plane that will allow effective use of
network resources and simplify network operations and provisioning.

This company was one of the first to introduce an optical switch, and now
it has a mature platform based on relatively old R&D results.  The
CoreDirector system features include distributed networking intelligence,
which enables network-wide provisioning and management as well as a
variety of protection capabilities.  The CoreDirector is 640 Gbit/s
non-blocking, multi-service platform.  Its smallest granularity of
switched bandwidth is STS-1.  The last version of this platform was able
to support standard BLSR/MS-Spring  architectures.  Its distinguishing
feature is software configurability for various SONET/SDH interface rates.
 Ciena also has a smaller version of the switch, CoreDirector CI.

This is the first half of the report.  The OEO section will be completed
in the next issue of "ViewsLetter on Provisioning," with the second
section on OOO switches.  A link will let you download the full paper from
    --William A. Flanagan, Editor and Publisher

If you have comments on this issue, share them by email to

"Flanagan Consulting" and "ViewsLetter" are Service Marks of W. A. Flanagan, Inc.
 Updated:  11 June  2003

Flanagan ConsultingSM
W. A. Flanagan, Inc.
45472 Holiday Drive, Dulles, VA 20166
Ph:  +1.703.242.8381
Fx:  +1.703.242.8391