Routed Optical Networks

[@James_Knott]

As an old telecom guy, going back to the analog days, I found your comments about going hop to hop to hop to cover a long distance curious. Even half a century ago, when I got started in the business, there would be trunks carrying the traffic between distant locations, without being broken down at each hop. These days, there could be some fibres used for short hops, but also others that are essentially passed all the way between distant points, without passing through routers routers along the way. And yes, they worry about power levels, as they did even back in the analog days, where too strong a signal could result in distortion and intermod between circuits. Even the test tone levels were reduced, when testing over analog microwave links, for that reason. BTW, there is also Coarse Wavelength Division Multiplex (CWDM) often used over short distances, where perhaps a dozen wavelengths are used over a fibre and the appropriate SPF modules are used for the wavelength. These could plug into any standard router that can accept SPFs. In this situation, the multi-wavelength would connect to a diffraction grating filter, which splits out the various wavelengths and then connects to the individual devices.

[@BarbaraAnne]

Because coherent optics can be plugged directly into routers, this video was more about whether using routers was sufficient or if it makes sense to retain an underlying ROADM network. I think it makes more sense to use an underlying ROADM network to continue to make use of router bypass - especially since there is more traffic that is being backhauled to the POPs and handed off. The "hop by hop" referred to pushing all traffic through the routers, which does not make as much sense to me in most instances. I appreciate your comments on the video.

[@cod06]

Interesting topic, I have had little experience with dwdm, mostly wdm, cwdm, for access we usually use gpon or gepon.

[@BarbaraAnne]

Thanks for sharing that. With coherent plugs and more fiber in place, the optical network is extending its reach and it does raise interesting network architecture questions.

[@dabb11]

Hello, I was under the impression that you seem to have come to as well using ROADM rings to actually do all the movement of data between large spread out aggregation boxes and having a few ROADM cards would let you put huge amounts of bandwidth onto few fiber pairs. Some vendors will do 32+ 1.2Tb/s channels in just the C band and another 32+ in the L band. Why is it that some vendors like Nokia and Cisco are really pushing for integrating Dense Wave Division Multiplexing Links straight into the routers and connecting them straight to each other arguing that the ROADM functionality can be bypassed entirely for operation expenses reasons since now you only manage "1 layer"? Routers are far far more expensive than optical transport equipment and offer far far less throughput per fiber run and just like in your example the closer you get to large peers or the internet drain you need higher and higher capacity chassis to switch traffic from downstream boxes.

[@BarbaraAnne]

That's what makes sense to me too. Also, from those statistics I quoted, more than 90% of web traffic is generated from the mobile network. All of that traffic has to be backhauled to the MSC so it makes sense to just forward that on the ROADM network as well.

[@cabrademonte]

Not only those vendors are pushing routed optical networks, but on aws net402 (deep dive on aws network infractsructure) presentation they said aws is doing routed optical network. I believe i also saw a document on microsoft doing the same. Obviously some orgs had found benefits for having the coherent optics on the router. I would also like to point out that these days, is the optic that cost the most. Back in the day router ports were expensive, but these days the cost of router port is not as much as they used to be. The cost has being moved to the optics.

[@pedrosuarez2349]

the electromatic spectrum used must go to gamma or delta......for security reasons.......the magnetic resonance of just color wavelength is easily detectable by its backward magnetic signature easily translated by a condenser that mainpulates the sound into patterns.....an optical morse code if you will......

[@dabb11]

@@pedrosuarez2349 I'm not sure what you're talking about, delta and gamma bands? I'm not a quantum optics guy but I know enough physics to get by for sure. In the NIR bands used in optical transceivers there are the O, E, S, C, and L bands. I don't see how (increasing?) the frequency of your baseband avoids any sort of waveform related security issues, the fundamental properties of the magnetic field would be totally unchanged. You only get half of the picture from the magnetic field anyways as polarization of the light field in modulation effects both the electric and magnetic fields, same with any sort of frequency modulation (phase is also effectively frequency modulation).

[@tridang4714]

Im a telco guy, you didnt include the evolution of PDH to SDH then NGSDH to CWDM to DWDM and now OTN. We actually have more optical transport on the local networks than long haul. The IP and Ethernet guys always believe that their P and PE routers are directly connected via fibre over a 10Gbe port which isnt always the case. ROADM isnt cheap because its design is multiplexed similar to SDH and you need to build a ring and cannot just do a hub and spoke as per the older passive DWDM networks for best practice. Its also not economical for enterprise transport because the client ethernet ports on muxponder transponders are typically clear channel and you still need a switch or router to do your traffic engineering and routing etc. Its mainly a transport multiplexing purpose as how we have deployed them since 1990 hah.