Skip to main content
Message Font: Serif | Sans-Serif
No. of Recommendations: 160
Adding Intelligence to Optical Networks

I have written this in two main parts. The first section dealing with why we need an intelligent network.
The second the companies involved.

1) Why Optics ?
2) Why people are buying and will still buy Sonet - Dumb optical networks
3) Sonet + DWDM - Fast and Dumb
4) So why Intelligent Optical Networks?
5) The Next Generation Nework
....c)Optical Switch

6) Intelligent Optical Companies
a) Ciena
b) Cisco
c) Nortel
d) Sycamore
e) Tellium

7) The original question I wanted to answer
8) The winner is?
9) Queries
10) Thoughts

1) Why optics?

Data traffic is doubling every 6 months.
This is beyond what can be handled by electronic means.
Only an optical network can handle these projected volumes.
High demand and expensive equipment will lead to obscene spending levels.

Optical Spending
RHK estimates...1999...........2003 ( in billions)

Of the $31 billion - Intelligent Optical networks potential is estimated to be worth around $10 Billion by 2003.

Pioneer's view

Contemporary networks are built to a standard known as SONET. The SONET standard has it's limitations but there will still be a lot of SONET equipment sold over the coming years.

2) Why people are buying and will still buy SONET - (Why buy dumb optical networks?)

Synchronous Optical NETwork SONET is the current optical standard in communication networks. SONET networks allow interoperability, high quality service and protection. While allowing unlimited data flow. SONET will not die just yet.

The SONET Legacy - Why did we have SONET?

First used in the '90's It is an agreed standard which avoids the mayhem of everyone having their own proprietary solution. The SONET standard makes this happen.

SONET is :

1) Built in rings - data travelling in the ring reserves the entire ring. Data passes between two points on the ring. The ring can then be used for a new batch of data. There is no restriction
2) Built for reliability - each ring has a backup ring which will only be called into action if the primary ring ( fibre) is broken.
3) Built at one speed - equipment on the ring can only handle data passed at 1 pre-set rate. This means a) If you want another speed you have to upgrade the entire ring b) Upgrading one ring causes more traffic which could mean upgrading adjacent rings etc.
4) Built for Voice - a data source gets a regular slot to travel on. Don't use it ? Then that slot is wasted. Using timeslots is a way of guaranteeing that conversations on the phone don't 'break up' delivering quality of service.
5) Built around Electric-Optical-Electric (O-E-O) conversions. To get data onto and off of a ring it goes through O-E-O conversions which are done by Add Drop Multiplexes (ADM). These are expensive in time and money.

It is important to stress the system must be reliable. Must. This is a vital ingredient in SONET and in any telecommunication network to come.

Note: the European equivalent to SONET is SDH.

SONET delivered a standard but the almost limitless capacity was delivered by DWDM.

3) SONET and DWDM - DWDM makes SONET Fast and Dumb

What is DWDM ?

DWDM allows many channels of light to run down a single fibre which previously only supported one. More data down the same pipe. However all this bandwidth abundance needs controlling. This is the next problem and the one which Intelligent optical networks address.

DWDM - tutorial from Lucent.

According to RHK the DWDM share is as follows

DWDM Market shares Nortel 37% , Lucent 29%, Ciena 14% , Pirelli 9% , Siemens 2% - Sycamore 2%.
DWDM bundled with Sonet Nortel 80%, Pirelli 7% , Lucent 6%, Siemens 3% , and Fujitsu 3%
DWDM not bundled Lucent 47% , Ciena 26%, Piirelli 13%, NEC 7% , and Sycamore 4%

4) So why Intelligent Optical Networks?

Great, why do we want intelligent networks?

We like SONET + DWDM because
1) Quality of service - the voice message must get through! It gives us protection
2) DWDM gives Speed! Speed! Speed!

We don't like SONET because
1) It forces us to use rings. The shortest way between two points is not 15 rings lying next to each other.
2) Too many conversions - every time we go from one ring to another we have to go through OEO - this means a) expensive b) more chances of error c) more complications on a complex network
3) No management - To allocate band width on a dumb SONET network can take 6 months. Yep you read that right. Think of all those rings. All needing to be manually checked that they've got free bandwidth on. Going to wait if you know someone with an intelligently managed network can put in a quote of 3 mouse clicks? No me neither.
4) No services - if we can't handle management you can forget the extras - as a minimum we expect dynamic band width delivery and flexible costing. While were here we want virtual private networks ( and b2b e-commerce will want extranets ( - an article on services - services the money - anything that makes putting services in place easier the better.

What are we looking for?

We need to keep
1) We still need quality of service - if it isn't reliable then our products won' t be bought

We want

1) Simpler upgrades. Every time we change speed. We need to upgrade the entire ring. All the ADM's ( These can cost $100,000's each.) And replace them with ADM's of the new speed. We want to avoid this slow and expensive.
2) Point and click allocation of bandwidth and services
3) A choice of architecture - we don't just want rings. Rings are an interim solution to give protection. If we have a mesh networks ( imagine lots of lines criss-crossing lines - lots of ways to get to one point from different routes). Then if there is a break we just re-route along one of the different route options. Less o-e-o's between 2 points. Cheaper and easier to expand. But we want companies to offer us an upgrade path from rings to meshes and support both.
4) Upgrade path for SONET and SDH systems.
5) Simplifying the network - ATM and SONET could be retired with their functionality built into IP over DWDM.

So what do we need for our improved network?

5) The Next Generation Network
We need as a minimum
a) Router
c) An optical Switch
d) Management software

5a) Router

While a missing router can be forgiven at the moment all the vendors are going to have to show that they have an integrated system where the router can interact with the switches to increase or decrease the load they are getting. This is needed to get the best quality of service. MPLS is an open standard but obviously this is going to be easier to tune if you control all the pieces.

It will be interesting to see if the router only companies have to show them interacting with other companies switches and DWDM devices seamlessly. Or will they make alliances ?

Provisioning has been demoed by Tellium, Avici

5b) DWDM
Yes please! However they will be getting more wavelengths down the same pipe. Which means more complexity and more need for management.

5c) Optical Switches ( for an Optical Network)

The current Add-Drop-Multiplexers and Cross-Connects only work at one speed. A fully optical switch does not care what speed we are piping down the fibres. Leading to faster and cheaper upgrades.

So this should be simple we want an optical switch.

Not that simple Fool! When is an Optical Switch not an optical switch ?

Sycamore and Ciena do not have an all optical switch. All the optical switches currently on the market convert from optical data to electric. This means that they will have a speed limit ( it's a lot of processing). A truly optical switch will just deflect the light on to another path. There is no limit to the data capacity of a fully optical switch.

In the near future someone will come out with a fully optical switch. Until them here's a current update.

I have tried to compare the two optical switch offerings from Sycamore and Ciena here. It gives an idea of the complexity involved in deciding what's best.

Comparison's of Sycamore and Ciena's Optical - Electrical - Optical Switches

Both Sycamore and Ciena's switches are a step up from using SONET equipment. notes
i) Switching - both Core Director and SN1600 can switch incoming data onto on to any outgoing line (256 X 256 lines at OC-48 or 64 X 64 lines at OC-192).
ii) grooming of connections. Grooming is when you break up a high speed link containing data for different locations and post them down the appropriate different lines.
iii) Be Cheaper - The switching and grooming functionality replaces the need for Add Drop Multiplexers and Digital cross connects. Simplifying the network and reducing space in the offices.
The Core Director at a typical configuration of $1-2 million is not cheap but it replaces even more expensive hard ware and saves space.
iv) Different switching protocols - from what I understand Ciena uses Ciena's proprietary protocol known as Optical Signalling and Routing Protocol ( OSRP). This is similar to ATM - ATM offers a good quality of service. Where as Sycamore uses IP (MPLS and OSPF) - using these open standards ( MPLS is being converged on by Cisco and Jumiper the leaders in high end routing ) seems to be the better choice. There is a caveat t that they are still working on quality of service in MPLS. So problems of this nature must be fixed before you can rely just on IP.
V) Upgradable - Sycamore is confident they have an upgrade strategy to an all optical switch. I do not know Ciena's position.

In conclusion - Ciena and Sycamore are offering simplified and cheaper switching functionality. Sycamore , from what I understand, has standardised round the more popular IP protocols - MPLS. While Ciena has a variant of ATM which offers a high quality of service needed currently but could be made obsolete.

Fully Optical and Optical-Electrical switches in harmony?

True optical routing, in which routers can read the addresses and other information carried in photons, is a long, long way away. "To switch packets in light would require a transducer that can convert light signals. Those are barely in the lab," says David Tolwinski, president and CEO of Tenor Networks Inc. (, an optical networking start-up.

If we accept that for the time being reading the address of photons isn't about to come in the near future. Then we would have a situation where fully optical switches and optical electrical switches are working together in an optical network. Optical switches will switch at the wavelength level ( i.e. can pick a wavelength out and send it on it's way) and the electrical switches will pick out the packets of information and route them as appropriate.

5d) Management - this means software

When you're buying a company like Sycamore or Ciena it's then the software is the real differentiation. There are plenty of companies out there with their switches or DWDM devices. However the bandwidth management device that handles fault detection , 1000's of streams of light and dosen't crash and is compatible with all you're optical devices is a compelling selling point. You can only do it with a kind telecom provider allowing you to learn on the job. Sycamore ( Williams) , Cerent ( QWest) and Ciena are getting time but other startups will have less time. Other companies will get less time if there is a working product on the market.

Can startups even afford to provide the software? Xros was bought by Nortel but it is a dumb switch.
The software might have to handle millions of wavelengths, and the slightest glitch could cause widespread service blackouts, he says. In other words, the software isn't something that could be dashed off quickly by a relatively small startup such as Xros.

Do we have any companies with an end to end solution?

6) The Companies

We want ideally to have a company that offers 1) Router 2) DWDM 3) Optical switch 4) Management Software

Of the companies I am looking at

Nortel and Cisco have all these.
Tellium could benefit from Lucent.
Ciena and Sycamore are missing the router.

So are they just me too? What's the depth to their products?

6a) Ciena

Recent History

Ciena first product was Dense Wave Division Multiplexing ( DWDM) - which enabled multiple light streams to pass down a single optic fibre.
(1998)Due to their small size and lack of product diversity they were going to be acquired by Tellabs however this broke down after ( among other things ) AT&T refused even to test Ciena 40 Channel DWDM system. Managment blaming Tellabs merger was taking up too much time and causing loss of revenue opportunites.
(1999 - March) Ciena acquired :
Lightea - products - optical switch and bandwidth management. The switch incorporates Add/Drop multiplexing and other functionality that clutters optical networks.
Omnia - products - product that sits between the core optical network and the electronic network. It packaged up a router and Add/Drop functionality into one box.
These additions were aimed at reposition Ciena into the larger space of Intelligent optical networks. Lightera's product became Core Director.
Omnia's product became Edge Director


The revenues are almost entirely from selling of the DWDM products and therefore not really relevant for looking for the winner in the intelligent optical network race. The Y/Y figure show the effect of being unable to compete with Lucent and Nortel during the 98 season and there subsequent recovery in the 99 season.

Revenues..........100.4 .....111.5.....128.8.....141.4.....152.2
q/q growth........10.2%......11.0%....15.5%....9.7%....7.7%
yr/yr growth......-25.2%...-21.9%....-0.2%....55.1%...51.6%

Intelligent Optical Network Revenues

Ciena announced a three-year contract with Williams Communications for its CoreDirector intelligent optical switch. Williams will deploy at least $40 million in CoreDirector systems over the next three years.

Market Share

DWDM - Ciena has 14% of the DWDM market ( behind Lucent and Nortel )

Consumer concentration - GTS Hermes, Sprint and WorldCom total almost half of the revenues.

Ciena Products

They have grouped all their products under the "Tool Box" - badge.

Core Director

Optical switch of the "O-E-O" variety - i.e. Electric core ( see switches below )


From the comments of the management. Edge Director has been a disappointment. Apparently due to a lack of functionality.

Core Stream

DWDM - Ciena have maintained a good range of DWDM products and 14% of the market share. Better than it sounds because half of the market for DWDM is "closed". You buy you're SONET gear and you get you're DWDM in the kit. This is almost exclusively Nortel's market.

Wave Watcher

Seems to be protection management.



CIENA manufactures on it's own facilities. Although this gives them stock control and allows them to take advantage of the current optics shortages they will eventually have to compete against the likes of JDSU where margins are going to be pushed.
I am not sure where Core Director is manufactured.


Core Director is being tested by 5 clients ( with a 6th one soon to be started or started )
Core Director has been deployed by Williams who have placed and order and iaxis and three others.

Ciena's Customers for other products include

Bell Atlantic, Bell South , Cable & Wireless, Enron, Williams Comms , Global Crossing iaxis.

6b) Cisco - The Optical Frankenstein

Cisco had only one part of the optical map the terrabit router


Cisco has entered the optical market buying the following.

Add-Drop-Multiplexer....Cerent - Cerent 454
Optical Switch.................Monterrey. 2000 Wavelength Router.

Cisco buy Cerent

Why Buy the Cerent 454
1) Reduce the number of Sonet components needed in the telecommunication network
2) Help the transition from a voice to a data network
3) 100 Carrier customers - big user base for their product

Cerent's - customers include Alltel , GST, Qwest Williams

Cisco buy Monterey

Monterey's 2000 Wavelength Router
-Optical switch- says it complements IP, MPLS
-eventually support 160 Tbps

The idea is that you plug the switch into the Core DWDM systems and leave the rest up to the switch.
The switch can handle restoration and Multiple level of service - Mission critical to bulk service.

Monterrey bandwidth control - WaRP , proprietary, 50 ms restoration
Pro - mesh network

Not good at grooming. As you can see from the press release. The router works with OC-48 and above

Monterey - Management

Jon Bayless, chairman of the board of Monterey and former chairman of Ciena Corp

Cisco buy Pirelli

Gave Cisco - DWDM
4th in Market.

Cisco and Corvis

Cisco has a 10% stake in Corvis. It often takes a stake before buying. Corvis looks like it is a player as well.

Corvis and

The Cisco strategy

They are late to the party but there still food on the table. Cisco can muscle in on the action with their excellent sales staff. With the addition of Pirelli they are in a position to be able to customise and get the various parts interacting. However, they will not have had time to produce an integrated strategy i.e. competitive advantage like, Sycamore has or the market dominance of Nortel. The wild card is Corvis. It seems to have a strategy. Will Cisco and Corvis marry?

6c) Nortel

Though dominant in the SONET area they were talking about their product line while Lucent, Alcatel, Tellabs and Sycamore were doing demos.

Nortel's solution was the Optera

Read all about Optera here.

Nortel buy Qtera

Qtera could affect Nortel's strategy

"Another potentially interesting aspect involves Qtera's philosophy concerning network restoration. The company had set itself the goal of incorporating dense wavelength-division multiplexing (DWDM) and restoration capabilities into the same optical platform. "

See - you need to log on but you just need to fill in a form

6d) Sycamore

Sycamore researches, develops and sells intelligent optical network products. Their products aim to add management and flexibility to the dumb SONET networks. Their product can either upgrade old SONET networks or develop a fresh cutting edge network.


With the build out still to really get going and with just a few months trading to evaluate the company a clear picture is difficult to come by. However Sycamore now has two contract which are multimillion and ratable over a number of years which will help income visibility.

Revenues................$11. 3..........$19.5..........$29.0
Revenue Increase .....................+73%.........+ 49%

Net loss for the second quarter of fiscal year 2000 was $1.6 million compared with a net loss of $3.1 million in the second quarter of 1999

Sycamore Products

I am only going to give an overview. The detail you will find on Sycamore's home page.

Silvx ONMS - for management of the optical networks. Consists of :
SilvixSource - Management on Sycamore's optical products ( all of the below)
SilvixManager - runs at the central office and gives precise bandwidth management and
other services.

SN6000 - Use to get many smaller (OC-48 connections) onto the big long haul connections (OC-192) - WDM and ADM functionality.
SN6000 is cheaper than SONET solutions and use less power
1) The process of fitting on smaller onto larger is known as multiplexing.
2) What does OC-X mean?

SN8000 - A more flexible SN6000 - takes OC-3 to OC-192 + has add-drop functionality. Use on long haul
and short haul ( metro ) networks.

SN16000 - "optical" switch - it converts from optical to electrical before switching. In the future there will
also be switches which do not convert into electrical


These guys and galls are not in any way stupid. For example Eric Swanson - Chief scientist and Co-founder was group leader of the Advanced Networks Group at MIT's Lincoln Laboratory. Or Richard Barry - was the chief network architect of the Advanced Networks Group (formerly know as the Optical Communications Technology Group) at MIT's Lincoln Laboratory.

Have a look

ODSI is a group organised by Sycamore with the vision to try to organise the optics players for interoperability. Although this has attracted the start ups and the smaller companies but resulted in a luke warm reception from the bigger player ( Nortel and Lucent are not ODSI members) it shows that Sycamore has some leadership and respect even if the initiative is a failure. While Ciena has become a member of it.



Sycamore follows the Cisco model. They hold only a few critical components in inventory and the rest of the manufacturing is by Celestica, Inc. This saves on working capital while avoiding unnecessary build out of the company structure.


Williams - $400 Million - spend $100 million a year for 4 years.
Note: Sycamore gave Williams stock before Williams bought into Sycamore

Qwest - building an multi-vendor optical network - Sycamore is in lab trials with Qwest

Iaxis - conducting field trials of the SN 16000 switch in France and Germany

Enron ( Jan 31st 2000) - SILVX - may be embedded into Enron's management system to look after the SCMR's products. Potentially worth up to $200 over 3 years.

LD COM ( Feb 14th 2000) - SN8000, SILVX -

Utfors (Feb 14th 2000) - - SN8000 , SN16000 , SILVX )

Storm Telecommunications Limited (March 17th) - deal worth $40 Million signed to execute Storm's o-commerce strategy

Sycamore are still unlocking shares ( written 24th March) to see how it's going


Optics companies cash in on the speed of light

6e) Tellium

Spun off from Bellcore in 1997. Tellium has made a deal with Lucent who will sell Telliums products along with Telliums.

The Aurora switch

The switch has 512 channels , 1280 Gbps capacity but gooming only begins at OC-48. Where as it does manage to use MPLS.

The management system

Break the light flow and see the correction!

I don't have a lot of information on it. But it seems it looks a player. It will be interesting to see if Lucent can sell it's products.

7) The original question I wanted to answer to my own satisfaction - Ciena V's Sycamore

Contract Won?

Core Director should produce revenues but the future of Edge Director is less certain.
Note: Ciena already had Enron and WilliamsSycamore as customers but
a) Won substantially larger contract than Ciena ( scmr - $400 over 4 years - cien $40 over 3 years )
b) Enron offered a substantial ( $200 ) contract. Ciena has not.

It will be interesting to see what happens with Iaxis. Another customer of Ciena but both Sycamore and Ciena are in trials with their systems.

8) The winner is?

Sycamore has the optical vision. They were the start up that first developed Intelligent optical networks. Ciena, Cisco and Nortel bought into it or were late starters. Tellium has optical vision but I am unaware of any contracts won.
Cerrent ( who may end up with Cisco ) is trying out at Qwest. We will see who is chosen.

So we can say it's all nicely posied. Sycamore has had a few deals. Ciena is an early bird but will it get the worm? Cisco and Nortel is prowling around. While Lucent and Tellium should make an interesting combo.

The big spenders are AT&T , World Com and QWest and these contracts have yet to be won.

I like Sycamore but will I put money on it. That's the question?

9) Queries?

Mesh Network needs interoperability

With the topology of the networks changing from rings to mesh networks and the fact that there's going to be 1000's of miles of fibre out there. There going to be interaction between networks. Will the companies agree to a standard? Will they just go there own way and muddle through. I would think that if I was an telecom service provider I would want the option to interact with other networks. Will they force an agreement?

Can quality of service be maintained with MPLS yet? Is it something that is being worked on at the moment.

9) Thoughts

I started this as a comparison of Ciena and Sycamore however it all got a little too involved.
Anyway I have tried to outline all the characters in the puzzle and as you might expect given my initial leaning with a little more detail on Ciena and Sycamore.

In addition I have cut short the section on Switching. This is complicated and would add a hideous amount of length to this ( already bloated ) posting. I may read about them more later and make a posting. Anyway with the developments coming a pace we may soon have the technology to load this posting quickly

Print the post  


What was Your Dumbest Investment?
Share it with us -- and learn from others' stories of flubs.
When Life Gives You Lemons
We all have had hardships and made poor decisions. The important thing is how we respond and grow. Read the story of a Fool who started from nothing, and looks to gain everything.
Contact Us
Contact Customer Service and other Fool departments here.
Work for Fools?
Winner of the Washingtonian great places to work, and Glassdoor #1 Company to Work For 2015! Have access to all of TMF's online and email products for FREE, and be paid for your contributions to TMF! Click the link and start your Fool career.