What are your experiences with alien waves, managed spectrum, spectrum as a service, etc? ----- Mike Hammett Intelligent Computing Solutions http://www.ics-il.com Midwest-IX http://www.midwest-ix.com
It’s one of those things that makes a lot more sense on paper than in practice. I’ve found it to be operationally difficult from the perspective the provider and the user, primarily but not solely because any “co-managed” system is going to lend itself to finger pointing when issues arise. Even if it makes more commercial sense at first blush to prefer a spectrum solution over dark or traditional waves, I suspect that factoring in “labor cost wasted over unproductive troubleshooting” changes the equation a bit. I also suspect that continued price compression on optical hardware will lead to fewer and fewer situations where it might make commercial sense at first blush too. YMMV of course and there may be reasons beyond simple commercial models where spectrum might make sense for you, but I’d urge you to only consider it doing with a provider where you’ve had a track record of operational success working with them. Dave Cohen craetdave@gmail.com
On May 12, 2024, at 2:17 PM, Mike Hammett <nanog@ics-il.net> wrote:
What are your experiences with alien waves, managed spectrum, spectrum as a service, etc?
----- Mike Hammett Intelligent Computing Solutions http://www.ics-il.com
Midwest-IX http://www.midwest-ix.com
On 5/12/24 20:35, Dave Cohen wrote:
It’s one of those things that makes a lot more sense on paper than in practice.
Not anymore. The majority of SDM subsea cables built with uncompensated fibre are using managed spectrum and spectrum sharing as viable business models for a not-so-insignificant population of their customer base.
I’ve found it to be operationally difficult from the perspective the provider and the user, primarily but not solely because any “co-managed” system is going to lend itself to finger pointing when issues arise.
And there is a case to be made for that concern. However, if you are in a position to be able to sell spectrum, it is very likely you are going to be implementing a vendor of note. Since that is most likely to be the case, those vendors have spectrum controllers that make this a viable business model, albeit at a $$ premium. This is not the type of service small DWDM operators using new-age DWDM vendors would typically be looking to sell. Such operators have to deal with keeping the lights on, never mind esoteric services like spectrum.
Even if it makes more commercial sense at first blush to prefer a spectrum solution over dark or traditional waves, I suspect that factoring in “labor cost wasted over unproductive troubleshooting” changes the equation a bit.
Alien wave and spectrum services attract a very high income, mainly through a capex-based upfront cost (IRU) that can be attractive to the host network. At those levels, providing their vendor has decent support for spectrum services, the revenue gain more-than makes up for all the logistical admin.
I also suspect that continued price compression on optical hardware will lead to fewer and fewer situations where it might make commercial sense at first blush too.
Well, legacy DWDM vendors will continue to charge a premium even for what is now standard electrical bandwidth services. Why? Because they still have all that legacy stuff to support, all their R&D to recoup, and because the bulk of their customers are no longer the telco, but content folk. New-age DWDM vendors are focused on coherent optical networks, which are primarily 100G and 400G. Why? Because that is where MSA and OpenROADM are currently at re: commercial availability. The legacy vendors will develop proprietary coherent pluggables that will support funky things such as 800G, 1.2T and 1.6T, but those won't be industry standard for some time (800G is getting there, though). What all this means is that if you are a legacy operator that is careful about spending money on newer DWDM technologies, a spectrum service from a larger carrier is going to be more attractive than ripping out your entire line system just so you can get from 10G or 40G to 400G. Of course, if you are a monopoly and have no alternatives to lean on, this doesn't count.
YMMV of course and there may be reasons beyond simple commercial models where spectrum might make sense for you, but I’d urge you to only consider it doing with a provider where you’ve had a track record of operational success working with them.
New-age DWDM vendors are not the workhorse of most of the large DWDM operator networks out there. That means that any operator of note you are likely to run into is going to be a Ciena, Infinera, Nokia, Adva, e.t.c., house, or something along those lines. Those vendors have reasonable spectrum-based solutions that smaller DWDM operators or ISP's would be willing to spend money on to avoid having to upgrade or deploy an entire line system. The reason that is feasible is because those larger operators are running vendors who push beyond what the MSA and OpenROADM groups are prescribing. You can already get coherent 100G and 400G channels on new-age DWDM vendors... that is not rocket science anymore. Of course, there is always the IPoDWDM question... but if I'm honest, I am still as unconvinced now in 2024 as I was back in 2008 that optical and IP folk would have a meeting of the minds on this. Mark.
Mark, Many/all of these points are fair. My experience is purely terrestrial and obviously both the capacity and economic calculations are vastly different in those situations, which I should have called out. However, I don’t think that the optical vendor is really the challenge - I would agree that, generally, spectrum is going to be available through larger providers that are using “traditional carrier grade” platforms - but rather at the service provider level. When something invariably breaks at 3 AM and the third shift Tier I NOC tech who hasn’t read the service playbook says “I don’t see any errors on your transponder, sorry, it’s not on our end” because they’re not aware that they actually don’t have access to the transponder and need to start looking elsewhere, that’s the sort of thing that creates systemic challenges for users regardless of whether the light is being shot across a Ciena 6500 or a Dave’s Box-o’-Lasers 1000. Dave Cohen craetdave@gmail.com
On May 12, 2024, at 5:34 PM, Mark Tinka <mark@tinka.africa> wrote:
On 5/12/24 20:35, Dave Cohen wrote: It’s one of those things that makes a lot more sense on paper than in practice.
Not anymore.
The majority of SDM subsea cables built with uncompensated fibre are using managed spectrum and spectrum sharing as viable business models for a not-so-insignificant population of their customer base.
I’ve found it to be operationally difficult from the perspective the provider and the user, primarily but not solely because any “co-managed” system is going to lend itself to finger pointing when issues arise.
And there is a case to be made for that concern. However, if you are in a position to be able to sell spectrum, it is very likely you are going to be implementing a vendor of note. Since that is most likely to be the case, those vendors have spectrum controllers that make this a viable business model, albeit at a $$ premium.
This is not the type of service small DWDM operators using new-age DWDM vendors would typically be looking to sell. Such operators have to deal with keeping the lights on, never mind esoteric services like spectrum.
Even if it makes more commercial sense at first blush to prefer a spectrum solution over dark or traditional waves, I suspect that factoring in “labor cost wasted over unproductive troubleshooting” changes the equation a bit.
Alien wave and spectrum services attract a very high income, mainly through a capex-based upfront cost (IRU) that can be attractive to the host network. At those levels, providing their vendor has decent support for spectrum services, the revenue gain more-than makes up for all the logistical admin.
I also suspect that continued price compression on optical hardware will lead to fewer and fewer situations where it might make commercial sense at first blush too.
Well, legacy DWDM vendors will continue to charge a premium even for what is now standard electrical bandwidth services. Why? Because they still have all that legacy stuff to support, all their R&D to recoup, and because the bulk of their customers are no longer the telco, but content folk.
New-age DWDM vendors are focused on coherent optical networks, which are primarily 100G and 400G. Why? Because that is where MSA and OpenROADM are currently at re: commercial availability. The legacy vendors will develop proprietary coherent pluggables that will support funky things such as 800G, 1.2T and 1.6T, but those won't be industry standard for some time (800G is getting there, though).
What all this means is that if you are a legacy operator that is careful about spending money on newer DWDM technologies, a spectrum service from a larger carrier is going to be more attractive than ripping out your entire line system just so you can get from 10G or 40G to 400G. Of course, if you are a monopoly and have no alternatives to lean on, this doesn't count.
YMMV of course and there may be reasons beyond simple commercial models where spectrum might make sense for you, but I’d urge you to only consider it doing with a provider where you’ve had a track record of operational success working with them.
New-age DWDM vendors are not the workhorse of most of the large DWDM operator networks out there. That means that any operator of note you are likely to run into is going to be a Ciena, Infinera, Nokia, Adva, e.t.c., house, or something along those lines. Those vendors have reasonable spectrum-based solutions that smaller DWDM operators or ISP's would be willing to spend money on to avoid having to upgrade or deploy an entire line system.
The reason that is feasible is because those larger operators are running vendors who push beyond what the MSA and OpenROADM groups are prescribing. You can already get coherent 100G and 400G channels on new-age DWDM vendors... that is not rocket science anymore.
Of course, there is always the IPoDWDM question... but if I'm honest, I am still as unconvinced now in 2024 as I was back in 2008 that optical and IP folk would have a meeting of the minds on this.
Mark.
On 5/13/24 00:11, Dave Cohen wrote:
Mark,
Many/all of these points are fair. My experience is purely terrestrial and obviously both the capacity and economic calculations are vastly different in those situations, which I should have called out.
Actually, terrestrial economics are easier to consider because you have the one thing the subsea applications don't have in abundance... power. Fair point, terrestrial revenues are significantly lower than subsea revenues on a per-bit basis, but so are the deployment costs. That evens out, somewhat.
However, I don’t think that the optical vendor is really the challenge - I would agree that, generally, spectrum is going to be available through larger providers that are using “traditional carrier grade” platforms - but rather at the service provider level. When something invariably breaks at 3 AM and the third shift Tier I NOC tech who hasn’t read the service playbook says “I don’t see any errors on your transponder, sorry, it’s not on our end” because they’re not aware that they actually don’t have access to the transponder and need to start looking elsewhere, that’s the sort of thing that creates systemic challenges for users regardless of whether the light is being shot across a Ciena 6500 or a Dave’s Box-o’-Lasers 1000.
I think you are contradicting yourself a bit, unless I misunderstand your point. Legacy vendors who have spectrum controllers have made this concern less of an issue. But then again, to be fair, adopting spectrum controllers along with bandwidth expansions via things like gridless line systems and C+L backbone architectures that make spectrum sales a lot more viable at scale do come at a hefty $$ premium. So I can understand that offering spectrum independent of spectrum controllers is going to be more trouble than it is worth. Ultimately, what I'm saying is that technologically, this is now a solved problem, for the most part. That said, I don't think it will be the majority of DWDM operators offering spectrum services en masse, for at least a few more years. So even if you want to procure managed spectrum or spectrum sharing, you are likely to come up against a limited set of providers willing to sell it, if at all. Mark.
On 5/12/24 14:08, Mike Hammett wrote:
What are your experiences with alien waves, managed spectrum, spectrum as a service, etc?
Your outcomes will vary depending on whether this is deployed for terrestrial or subsea networks. Subsea networks don't typically do alien waves, but rather, managed spectrum or spectrum sharing. This is especially the case on the newer SDM-based uncompensated submarine cable systems, where there is a huge volume of fibre pairs that makes this feasible, if not the most economical way to sell the asset to volume customers. For terrestrial, alien waves were the original model, and in my opinion, the preferred one, because all the host network has to do is provide a port on their filter with a wavelength. The filter isolates the adjacent signals from one another, which improves launch OSNR. That said, managed spectrum and spectrum sharing are quickly replacing alien waves as the preferred deployment option for terrestrial networks, which can largely be blamed on advances for the same happening on the submarine side of things, even though the original idea was mostly driven by GEANT and a bunch of European NREN's back in the day. Managed spectrum and spectrum sharing are more problematic because the chance of broadcasting bad noise to all other channels increases. Yes, major DWDM vendors now do have significantly improved optical power management systems (a spectrum controller, let's say) that will interact with the WSS in their ROADM, where the ROADM will set the centre frequency and its width, which helps to restrict any negative impact to launch insertion, and not toward the line side. Different vendors will have different spectrum controller options that make managed spectrum and spectrum sharing services either simple or difficult to deliver on their specific type of gear. If it is something you want to be serious about, this will be the one time where PoC'ing all the vendors you are interested in is worth your time. It would also be useful to understand how each vendor supports things such as T-API (Transport-API) and other OpenROADM open architecture features to improve wavelength and optical power management characteristics between different vendors sharing a single OLS (Optical Line System). You may find that support for T-API and other OpenROADM standards may be spotty to non-existent with many vendors, but a vendor with a solid roadmap is certainly not a waste of your time. Major traditional vendors like Ciena, Infinera, Nokia, Adva, Ribbon, and such, will have very extensive spectrum controllers, but they will come with the requisite $$ premium. Newer vendors whose platforms are based primarily on coherent pluggables approved by the MSA and OpenROADM will support alien waves, but may struggle to offer a comparable spectrum controller solution for managed spectrum and spectrum sharing, even if they may have a rudimentary ability to do so. Due diligence is highly warranted here, as the landscape is changing on a daily basis. In essence, "virtual fibre pair services" (if I can call them that) is a matter of security, by way of total optical power control. What you want the vendors you consider to answer is: * If a spectrum customer erroneously provisions spectrum outside of their allocated bandwidth, how does the host network deal with that so that it does not impact any other spectrum customers on the same fibre pair? * How do you effectively restrict spectrum customers from only being able to access just their allocated spectrum, where a simple broadband splitter would not be sufficient for this? * How do you monitor the optical spectrum between each spectrum customer to ensure optimal optical performance on a per-spectrum-customer basis? * Especially for subsea applications, but nowadays, also for terrestrial ones; how do you monitor and manage optical power requirements for unallocated spectrum, including previously-allocated spectrum to a spectrum customer whose signal has now "disappeared" due to a failure of their own SLTE (Submarine Line Terminating Equipment) or transponder? In other words, ASE (Amplified Spontaneous Emission) noise loading capability. Answering these questions makes it easier for interested parties looking to move away from procuring electrical bandwidth to, rather, procuring optical spectrum. Hope this helps. Mark.
Hello, Depends on the equipment but works just fine. I have tested Ciena Waveserver/AI and Ekinops equipment as alien wave into an existing Cyan ROADM system without a problem. Usually just make sure your TX power is 0 or +5 + whatever the minimum for the ROAM system to pick it up. On Sun, May 12, 2024 at 9:37 AM Mike Hammett <nanog@ics-il.net> wrote:
What are your experiences with alien waves, managed spectrum, spectrum as a service, etc?
----- Mike Hammett Intelligent Computing Solutions http://www.ics-il.com
Midwest-IX http://www.midwest-ix.com
participants (4)
-
Dave Cohen
-
eric c
-
Mark Tinka
-
Mike Hammett