Operators Hang Big Hopes on ROADMs

As reconfigurable optical add-drop multiplexers (ROADMs) show the potential to get more sophisticated, carriers are amping up their ambitions for a more flexible optical network.

The products and architectures to create that network were the subject of yesterday's Light Reading Webinar, "The Need for Next-Generation ROADM Networks."

The technologies, some of which are still emerging, add up to a more flexible optical network, one that can be remotely reconfigured on the fly. The goal is to keep traffic in the optical domain for as long as possible, said Webinar speaker Glenn Wellbrock, director of backbone architectures for Verizon Communications Inc. (NYSE: VZ).

Lower-layer transport tends to be less expensive, so a network that takes better advantage of the optical layer, as opposed to switching and routing, could have lower capital and operational costs. Moreover, now that ROADMs are ensconced in the network, operators find themselves wishing the devices could be more automated -- a dream that, admittedly, isn't new.

"Full reconfigurability has been on the operator wish list from day one, the early part of this decade, but it's taken time for the technology to mature to the point to make this actually possible in the network," Heavy Reading analyst Sterling Perrin said in his introductory remarks.

Operators want ROADMs to be colorless (delivering any wavelength to any port), directionless (allowing ports to send traffic in any direction on the network), and contentionless (allowing wavelengths to get reused by different ports in a node). Speakers on the Webinar outlined those requirements, which are apparently abbreviated as CDC these days.

Wellbrock also discussed gridless -- or, more properly, flexible-bandwidth -- ROADMs. The idea here is to mix wavelengths that use different widths of spectrum (some on a 50GHz grid and some on a 100GHz grid, for instance) or could handle wavelengths that don't conform to the International Telecommunication Union (ITU) grid at all. (See ROADMs Get Ready to Go Off-Grid.)

To make CDC and gridless ROADMs work, though, network management systems would have to step up, Wellbrock said. These optical capabilities open the possibility of accidentally sending the same wavelength down the same path twice, or of asking for a wavelength that would eat more spectrum than a fiber can spare.

"The network element's going to have to tell me no even if I hit the button three times," Wellbrock said.

A replay of "The Need for Next-Generation ROADM Networks" will be available on the Webinar archives page starting early next week, through September 2011.

For the next few months, you can also follow this topic at our Next-Generation ROADM Briefing Center at http://www.lightreading.com/roadm.

— Craig Matsumoto, West Coast Editor, Light Reading

Page 1 / 2   >   >>
fiber_r_us 12/5/2012 | 4:24:15 PM
re: Operators Hang Big Hopes on ROADMs

Of course, the operators also want it to be costless - This has what has held-up deployment of ROADMs.

Sterling Perrin 12/5/2012 | 4:24:15 PM
re: Operators Hang Big Hopes on ROADMs

Colorless, directionless, contentionless, and costless (CDCC). I like it!


paolo.franzoi 12/5/2012 | 4:24:12 PM
re: Operators Hang Big Hopes on ROADMs


Unless you are on a fixed wavelength WDM system, pretty much all this stuff is a ROADM.  ROADMs in that vein are about 10 or more years old.  What I thought we were now talking about were P-OTS where ethernet switching and some version of control plane is integrated into the ROADM.




fiber_r_us 12/5/2012 | 4:24:12 PM
re: Operators Hang Big Hopes on ROADMs

I know of very few carriers that are not already keeping their packet traffic in the optical domain as much as possible.  Why would a packet network engineer do anything else?  Packet networks are generally optimised to meet traffic demand and operators have been "expressing" packet traffic (i.e. bypassing routers as neccessary) for well over a decade.  Any time the traffic demand between two routers is sufficiently high, operators allocate a wavelength and connect the routers directly at the optical layer to provide for the capacity.

This function does not (and has never) required ROADM capability.  Simply allocating a wavelength through the network - whose traffic is not likely to move around (i.e. is not very dynamic) is all that is required.

Why does the industry associate "router bypass" with ROADMs?

Sterling Perrin 12/5/2012 | 4:24:09 PM
re: Operators Hang Big Hopes on ROADMs

I see a couple of things going on here that are related but also separate.

One trend is toward the integration of optical switching (ROADMs) and electrical layer switching into a single device - which is the P-OTx evolution.

The other trend is driving some functions that were previously done in the electrical layer down into the optical layer - because the optical layer is less costly in both capex and opex (and the opex is a big deal now). This could happen within a P-OT type of box, or the ROADM devices and the electrical switching devices could be kept separate.

One problem with ROADMs to date is that they have only been able to express traffic at intermediate nodes and any add/drop nodes needed a technician (so it was really only partial automation). With a fully-reconfigurable ROADM network, some operators see the ability to reduce the total number of OEO conversions required for transport. Other operators want to reduce the amount of traffic that touches routers during transport and are happy with OEO conversions as along as they occur at Layer 1 (using OTN). So, i think there is a bit of competition potentially betweeen a fully reconfigurable ROADM network and an OTN switched network.



fiber_r_us 12/5/2012 | 4:24:08 PM
re: Operators Hang Big Hopes on ROADMs

So, i think there is a bit of competition potentially betweeen a fully reconfigurable ROADM network and an OTN switched network.


Exactly.  The premise that ROADMs or P-OTS somehow enable "router bypass" that couldn't be (or hasn't already been) done with just with allocation end-to-end wavelengths seems like a red herring.  Operators have already bypassed routers at the optical layer - POTS/ROADM architectures are not going to improve this.

Pete Baldwin 12/5/2012 | 4:24:07 PM
re: Operators Hang Big Hopes on ROADMs

> Why does the industry associate "router bypass" with ROADMs?

They don't, necessarily.  I think this is more about Verizon wanting ROADMs to help it do router bypass more easily and more cheaply.

Maybe the optical-layer quote was the wrong one to use, because as Sterling points out, it overlaps into Verizon's super-POTS vision (which is about increasing the amount of router bypass).  Which is related, but this webinar was more about the superpowers that carriers want ROADMs to acquire.

paolo.franzoi 12/5/2012 | 4:24:06 PM
re: Operators Hang Big Hopes on ROADMs



I think the theory is that you could build (and I am really simplifying) a 100G ring and do sub-multiplexing via Ethernet at the drops at 10G.  This allows the backbone to have oversubscribed wavelengths.

Good idea?  I am not a transport guy, but I think that is the theory.



fiber_r_us 12/5/2012 | 4:24:06 PM
re: Operators Hang Big Hopes on ROADMs

The cheapest capex way to perform router bypass would be to express a static wavelength between the router pairs with traffic demand while passing through the minimum quantity of optical equipment possible.  This is what every operator currently does to optimise their router networks in order to minimise traffic transiting routers unnecessarily.  ROADMs, POTS, or any other additional kit in the middle of optical paths between routers won't make it cheaper.

The ROADM piece almost completely sounds like an OPEX provisioning play - that is, the operations people wouldn't have to "touch" as much optical stuff to provision a wavelength.  Yet, wavelengths between routers are fairly static - traffic patterns are not that dynamic in the operators' packet networks.  Thus, once an express wavelength between routers is provisioned, it tends not to move.  I know of one operator that has never de-provisioned a 10G wavelength between routers once it was there - only added more 10G wavelengths as traffic demand increased.

I also don't see how inserting POTS kit (layer 2 packet switching of some form) between routers and the optical layer makes anything cheaper - really reminds me of the ATM days - just replacing ATM switches with POTS switches, thus re-adding the layer we got rid of when ATM died.

Is Verizon saying their router people have built a poor network and have not been optimising it by expressing wavelengths between routers as traffic demands?  If this is the problem, maybe they need new router people!  If the router people *have* been optimising their topology and leveraging "express" wavelengths to bypass routers (which I suspect they are smart enough to do), then where do any savings come from by adding ROADM or POTS?

I understand why the optical transport world would like to have ROADMs (with all the features), as this would make provisioning in that part of the network more automated.  But, this has little to nothing to do with router bypass.  It also illustrates why it has taken so long for the industry to adopt: for the operator to spend money on ROADMs there must be value.  Router bypass is clearly *not* the value.  So, the operators must fall back on operational costs – always a difficult proposition – and clearly illustrates why operators would like ROADMs to be “costless” – as then it would become a “no brainer” to justify the operational simplicity that ROADMs promise.

Operators also try to make the operational costs appear high by claiming that there is massive churn in the wavelength topology.  Not sure where this would come from as neither customer nor inter-router 10G/40G wavelengths move much even over multi-year periods.

fiber_r_us 12/5/2012 | 4:24:05 PM
re: Operators Hang Big Hopes on ROADMs

Yes.  I have heard that idea - but, that presumes that the routers are somehow not good at muxing themselves or are somehow underutilising the 10G wavelengths that are allocated to them. This doesn't really decribe much of the infrastrucutre I have seen at the major carriers around the world.  If anything, the traffic growth has driven them to continue to add nx10G, nx40G (and soon 100G) capacity between routers.  There is not much room for sub-multiplexing.

Perhaps in the peripheral reaches of the network where capacity demands are far below even a single 10G you could consider this.  But, that doesn't seem like a big market.

Page 1 / 2   >   >>
Sign In