x
Optical/IP

Poll Paints Bright Future for CWDM

Coarse wavelength-division multiplexing (CWDM) could eventually overtake DWDM in metro markets, according to a Light Reading Research Poll on the subject staged in October.

A total of 249 people took the poll, and -- surprisingly -- 43 percent of them went so far as to say that CWDM would overtake DWDM in metro markets. Another 35 percent said CWDM would be widely deployed, but not as widely deployed as DWDM.

The full results of the poll can be seen here.

Until recently, CWDM was considered something of a poor cousin to DWDM -- partly because it could pump only up to eight channels down a pair of fibers, and partly because it wouldn't work over distances of more than about 50 kilometers.

Now sentiment appears to have changed, mainly because CWDM promises to help carriers out of a jam by boosting bandwidths in metro networks without busting the bank.

Opinions vary over how much carriers can actually save with the technology. Some CWDM equipment vendors say they can cut costs by a factor of three, while DWDM vendors say savings rarely exceed 20 percent. In the Light Reading poll, 30 percent of respondents thought CWDM could deliver savings of more than 50 percent compared to DWDM. Only 25 percent of respondents thought savings would be 20 percent or less.

Confidence in CWDM as a technology appears to have grown since a 16-channel version was standardized by the International Telecommunication Union, Standardization Sector (ITU-T) last June. In the Light Reading poll, 82 percent of respondents said CWDM was mature enough to be used in metro networks; only 11 percent thought it wasn't.

The other reason for CWDM's increasing popularity is probably the growing availability of low water peak fiber, required for 16-channel systems, and the arrival of innovative CWDM solutions from the likes of Cisco Systems Inc. (Nasdaq: CSCO) and Transmode Systems AB (see Cisco Adds CWDM to Switches and Transmode Makes a Little Go a Long Way).

These and other issues concerning CWDM will be put in the spotlight in a Light Reading Webinar scheduled for next Tuesday, November 19, and hosted by Yours Truly. To get more information and to register for the Webinar, entitled "CWDM: Optical Capacity Without the Cost", please click here.

— Peter Heywood, Founding Editor, Light Reading
www.lightreading.com
Page 1 / 4   >   >>
dwdm2 12/4/2012 | 9:19:07 PM
re: Poll Paints Bright Future for CWDM gea said: "Nah. Just for 16 wavelength CWDM. 8 Wavelength CWDM works just fine on regular SMF 28, which is why it's already being deployed."

Gea, what I hear you saying is: only 8 channel cwdm can be used over the exisitng fiber. For anything higher, one needs zwp fiber to avoid the mid zone loss peak. So this does not address the scalability issue over the regular, already laid down fibers that were being discussed on this thread a little ago.

Now as for the economics model etc., I have seen lots of presentations on how the DWDM is the best solution of the both world.

As I said before (and also mentioned by others), both CWDM and DWDM have their niches. None of these technologies are gonna go away in favor of the other.

BTW, it seems like Peter Haywood is a party to CWDM! Anyone knows why?

Cheers.
gea 12/4/2012 | 9:19:14 PM
re: Poll Paints Bright Future for CWDM "BUT one must first lay down this ZWP fiber for CWDM to work!"

Nah. Just for 16 wavelength CWDM. 8 Wavelength CWDM works just fine on regular SMF 28, which is why it's already being deployed.
dwdm2 12/4/2012 | 9:19:14 PM
re: Poll Paints Bright Future for CWDM So the summary looks something like this:

CWDM is cheaper
Less initial investment
Less incremental investment
Good economic models
etc. etc. but,

BUT one must first lay down this ZWP fiber for CWDM to work!

So the question is: who gains the most from the "CWDM is better" argument?
whyiswhy 12/4/2012 | 9:20:01 PM
re: Poll Paints Bright Future for CWDM Great to see some posters who remember why it happened. The only thing I can add is that we see corporations snapping up (overbuilt/dark) fiber at bargain basement prices. RBOCS too are lighting a (single DWDM) wavelength/fiber at a time.

This will last until the existing glut of fiber is exhausted (point by point), and no longer. The bulk economics of DWDM cannot be beat lacking special circumstances like (virtually) free fiber. The price differences between a CWDM and a DWDM capable EDFA are small. Rights of way and installation costs are about the same as they ever were.

CWDM will find a long life and high volumes in enterprise / desktops at (sub?) five dollars per component in plastic.

JMHO

-Why

pigglywiggly 12/4/2012 | 9:20:13 PM
re: Poll Paints Bright Future for CWDM To MrLight's question:
"Out of interest is there any deployed products (not trial pleas) that use all16 wavelengths per the ITU GǣCWDM? "

Only 3 manufacturers that I know of have commercially released 16 wavelength systems, and I know of no deployments (with 16 wavelengths) of any of them:
Transmode
MRV
PadTec - Brazilian-based
lightmaster 12/4/2012 | 9:20:13 PM
re: Poll Paints Bright Future for CWDM Comments to a couple of things in the last post:

"the major reason (for DWDM) was by no means fiber exhaust but the elimination of SONET regenerators via EDFAs, and then the "leveraging" of the EDFA via WDM."

Granted that most of the savings comes from elimination of regens, there were major fiber exhaust problems that drove the adoption of the technology at Sprint. Sprint was a very early user of fiber and therefore had small fiber counts in their bundles. That, combined with a move to 4 fiber BLSR, gave them major fiber headaches. Unless they had a gun to their heads, they would have never taken the risk of buying from a startup like Ciena. The reasons you gave accounted for the majority of follow on sales of DWDM to new carriers after they became comfortable with the technology.

Regarding Metro, there are point fiber exhaust issues as well, driven by sections of fibers where multiple rings overlap. These require point solutions (not DWDM networks). RBOCs are already using DWDM in these applications and may opt for CWDM if it makes sense.

Agreed that in most cases the business case is CWDM versus fiber, because in most cases DWDM is not in the running due to cost.
gea 12/4/2012 | 9:20:15 PM
re: Poll Paints Bright Future for CWDM Sorry soup...like a lot of newcomers to the Optical Networking industry, you have overlooked THE critical question that ha to be answered for Metro DWDM systems: Why deploy WDM at all?

Contrary to the hype that was spewed all over the place as DWDM zoomed into long haul, the major reason was by no means fiber exhaust but the elimination of SONET regenerators via EDFAs, and then the "leveraging" of the EDFA via WDM. Thus, it was a no-brainer to deploy DWDM in the long haul even if you had plenty of dark fiber in the ground.

The same is not true of Metro systems. In the metro, OFAs are rarely needed. In addition, there's now TONS of dark fiber that was laid in response to the general "were running out of fiber!" panic in the 90s. So why does anybody need to deploy WDM at all, when they can put all those client signals over separate dark fibers, and save lots of costs?

The answer is basically that there really is no need for WDM in the Metro right now, and the fact that Metro WDM sales are almost nada is a strong indicator.

CWDM, however, can in some cases make a compelling case for WDM deployment, even when there's plenty of dark fiber lying around. But this only happens at low enough cost points (ie, where the cost of maintaining dark fiber is higher than CWDM). And it is in precisely those situations (no need for EDFAs, plenty of dark fiber) where channel count is a nonissue.

In addition, I still view the primary application of CWDM as the upgrade of corporate MANs. In this environment, there's often no need for transponders, thus making the desirability of CWDM MUCH higher than DWDM from a cost/complexity perspective.

As for AON, are you referring to Automatically Switched Optical Networking (aka ASON), or "All Optical Networking"? If you mean the former, I agree that's eventually going to be a killer "app". If you mean the latter, it's dead in the water, AND there's probably little need for such a thing now, given the state of off-the-shelf silicon available. (I worked on Bellcore's MONET project, and that's enough to make you think twice about all-optical networking.)
MrLight 12/4/2012 | 9:20:16 PM
re: Poll Paints Bright Future for CWDM In response to Peter Heywood's post 16"My understanding is that Nortel has rather a strange version of CWDM. In essence, it's Nortel's DWDM platform with most of the channels disabled, so that the remaining channels are spaced a long way apart - a bit like a comb with most of the teeth missing. I've yet to confirm this with Nortel."
---------

I am assuming you are referring to the OPTera 5100. One of the configurations is at you described, meaning DWDM lasers but CWDM-like filters based on just their band filters i.e. no DWDM channel filters. The advantage is that 5200 customers can re-use the 5200 cards in the 5100 shelf and therefore reduce their circuit card inventory while benefiting from a substantial cost savings of a cheaper filter arrangement (what Nortel calls the OMX cards). The other advantage is that NortelGs approach can run over standard SMF-28 fiber by operating in the 1530 to 1605 nm C & L bands and can if desired be upgraded to DWDM by replacing the filters with DWDM filter cards.

The other option for the OPTera 5100 is to use both CWDM laser and CWDM filters, incurring the associated inventory impact and new fiber requirements of operating in the S-Band when running 8 or 16 20nm CWDM channels.

If you look at the cost savings Peter, the biggest savings with CWDM telecom boxes is in the filters and not the lasers.

Now in regards to RGreg's post 20 "Yes, they definitely do that on their long haul OPTera platform. I've seen it installed like that in a couple of carrier hotels."

That is not CWDM, and its is not advertised as such, that is just the standard technique for equipping DWDM wavelengths, meaning you always under-equip each terminal to give you the option to add wavelengths to not just the fiber but to the site.

In addition another point that needs clarification is that the term CWDM when it started to be regularly used in 1997 did not cover the S-band, as per pigglywigglyGs post 21 statement Gǣrecently cited 16 wavelength capability of CWDM, as "standardized" by the ITU,Gǥ which may lend to the confusion about CWDM.

Out of interest is there any deployed products (not trial pleas) that use all16 wavelengths per the ITU GǣCWDM? The ONI ONLINE edge for example employs up to eight, 20 nm CWDM spaced wavelengths so it isnGt a 16 CWDM product. See http://www.ciena.com/products/... .

So will CWDM overtake DWDM?

Well for a lot of short-reach DWDM-like applications probably yes, but hopefully the standardization of CWDM gridding will also create new applications where DWDM was not practical.

But the real obstacle for CWDM to overcome is not DWDM's head start, but CWDM's difficulty in working into the existing fiber plant and the lack of tunable CWDM lasers.

MrLight :-)
lightmaster 12/4/2012 | 9:20:16 PM
re: Poll Paints Bright Future for CWDM A lot of the pro/con arguements here mix too many applications. Metro SP's certainly want things like 300KM rings, but that's not the application for CWDM.

CWDM is not a good "networking" technology. No add/drop. No amplifiers. It serves to provide more bandwidth over a sinle segment of fiber. In many cases, that's OK. Because of multiple rings that overlap routes, there are many cases that exhibit fiber exhaust only in one or more segments of the ring. In those cases, it's not desireable to put in an entire DWDM ring, but to solve the exhaust issue on the segment in question.

The issue is purely of cost comparisons. If CWDM is only incrementally cheaper, it will cost more in the end in terms of using two systmes where one could do the job, and the channel limitations will be a sticking point. If it is ten times cheaper, the other objections become easier to defend against.

I keep seeing differing numbers on the comparisons. Both DWDM and CWDM vendors show numbers that defend their positions. It's hard to tell unless you are a carrier getting bids.

Regarding the Light Reading poll, it portrays the opinions of the masses, which are seldom right. I wonder what it would have said two years ago about the death of ATM, Metro DWDM and packet rings killing SONET, 40 gig, the all-optical network, etc.

Still, I guess it beats an RHK or Yankee group forecast ;)
Soup 12/4/2012 | 9:20:17 PM
re: Poll Paints Bright Future for CWDM OK, let's have some fun with this:

My gut reaction to all of this makes me envision people of the dark (fiber) ages jumping up and down insisting that the world (of data rate demand) is flat. Hey, maybe it is.

CWDM as a long-term winner . . . not likely.
CWDM as a short-term winner . . . in some cases.

Word on the street is that Metro guys are a little nervous these days unless they can go 300 km or more with their systems, so why are some folks talking about 80 km as acceptable Metro systems? Also, I hear progress is all about more for less; it seems like CWDM is to a hacksaw what DWDM is to a Sawzall. Hacksaws are cheaper initially, but then they're just too darn slow and tiring to use.
The rules are simple:
Cheaper
Faster
Smaller

Cheaper: What's the cost per bit across the whole usable band in CWDM vs DWDM (for example, 16 CWDM channels vs 100 DWDM channels [50 GHz channel spacing, C-band])? I know, lots of other costs like more expensive lasers, mux/demuxs, etc, for DWDM, but has anyone added it ALL up objectively?

Faster: Duh, 40 Gb/s CWDM vs 1 Tb/s for DWDM? Things that don't make you go "hmmmm." (Oh, but do people NEED 1 Tb/s in Metro? I don't know, let's ask Bill Gates if he thinks people still only need 8 kB of memory in their computers).

Smaller: Again, per bit. I hear rack space costs money. How would the overall sizes per bit compare? And how logistic is it to hybridize CWDM to DWDM after you outgrow CWDM, and can you recover the cost in the useful lifespan of a CWDM system (what's that, maybe 3 years)? I don't know about all the stuff involved to "make it happen" for Metro, but I'd love to hear about from an objective source.

Then there's the dream of an AON, where Metro and Long Haul can mesh seamlessly. Why buy a slow, clumsy system that doesn't interface to the world around it (I know, there's protocols and such to worry about too, so it's not so simple)? No doubt, there are many dimensions to this problem (lucky for the porn-site users, there are also lots of smart people working on it). The best thing to do is try to stay out of local minima. Leave it to the goof-balls to mix up everyone's emotions to help drive them to the global minimum.
Actually, it doesn't matter to me what people choose. I just have to giggle when I read stuff like, "CWDM will crush DWDM." Remember The Kids in the Hall? The bit where the guy went around visually pinching distant objects between his fingers saying, "I crush you now!" C'mon.
Page 1 / 4   >   >>
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE