Top 5 Most Common Routed Optical Networking Myths
In this article
Recently I attended Cisco Live in Las Vegas, NV. As usual, the experience is always a humbling one, to say the least. Whenever I attend, I am awe-struck at the wealth of knowledge, innovation, effort and investment Cisco puts into this event each year. I was lucky enough to acquire a full event pass which includes a plethora of class sessions to choose from. These sessions are designed and orientated around a wide variety of topics. As an optical transport technical solutions architect, my top choices are generally around similar subjects. This year the event was packed with sessions around innovations in 400G routed optical networks with classes such as:
- Growing the Network with 400 Gbps Coherent Pluggable Optics
- Management of IP+Optical Networks Using an SDN Controller Architecture
- Improve Your Network Efficiency with 400G Optics
- The Journey Towards Routed Optical Networking
- Optical and Transport Innovations
- Coherent Pluggables: Revolutionizing Optical Infrastructure
- Designing Routed Optical Networks – IP/MPLS Considerations
- Interactive Discussion on Traditional Optical vs. Routed Optical Networking architectures
Here at WWT, we have spoken about routed optical networking through many articles, videos and communities. For a quick look at the entirety of our routed optical networking material, go to our community page on converged optical & packet to see the content catalog. We won't go in-depth on this topic during this article but for a general refresher, routed optical networking is basically extracting the coherent DWDM interface out of the traditional transponder or muxponder and placing it into a pluggable optic directly in the router. This topic is by no means new, as many vendors have implemented these exact strategies throughout the past decade. However, it has not been until recently that the technology, footprint and price have reached a point where the convergence of optical and IP is realistic. With this solution now attainable and cost-effective for many network operators, everyone is gathering information, education and misconceptions regarding the solution architecture.
I will be honest and say we at WWT get a front line to the new and innovative strategies leading OEMs like Cisco are deploying, planning and even forecasting for the next generation of solutions. This puts us in a unique position when attending sessions during Cisco Live. We can be hyper-focused on the questions during sessions and get a general vibe of how the community absorbs the new solutions. In this case, routed optical networking. From this perspective, we can gather some common misunderstandings around a particular product or solution.
Here are some common misconceptions I experienced while at Cisco Live 2022 around routed optical networking.
We will preface every question by saying there is no one size fits all routed optical network architecture. It is built to meet the operators existing and future demands within reason. With that out of the way, DWDM is a means to expand the site to site outside plant fiber capacity. Outside plant fiber is a costly, time-consuming commodity that shouldn't be consumed for each service added to the network. Especially if there are large distances between each location, we can expand the capability by adding DWDM to that precious commodity of outside plant fiber. A simple DWDM filtered solution combining multiple wavelengths onto a fiber pair can help address existing and future needs. This way, if an operator experiences times of exceeded growth, we do not need to burn another pair of outside plant fiber between locations. DWDM filtered solutions can be very cost-effective and, dare I say, a "cheap" solution to allow an operator to future-proof the network. DWDM filtered solutions come in sizes ranging from 2 channels up to 96. If distances are too long to accommodate a DWDM filter, then a DWDM amplifier can be used to boost the optical signals and reach expanded distances. For more information on passive DWDM filtered solutions, see our converged optical & packet community page.
Question 2: "I heard ROADM is no longer needed. Does this mean we should tear out our existing ROADM system?"
Again, no one size fits all routed optical network solution here. ROADM is still a very much viable and realistic solution for many network operators out there. ROADM systems are still being developed, expanded upon and invested heavily by all leading optical OEMs. Most modern ROADM systems today leverage Flex Spectrum capabilities. With Flex Spectrum ROADM networks, we can inject 400G signals and transport those in various fashions around the system. If the signal needs to be dropped from site to site, that is possible, much like a DWDM filtered solution. However, with Flex Spectrum ROADM, we can take those signals through locations leveraging passthrough capabilities allowing the operator only to drop 400G services where needed. This is an essential strategy for many network operators as it enables them to plan for future services that may not be realized initially. For more information on passive DWDM filtered solutions, see our converged optical & packet community page.
Question 3: "Is it true that 400G QSFP-DD ZR/ZR+ optics can only go 120 km before you have to regenerate?"
I will preface this question by saying that distance is significant, but more importantly, we need to consider the optical loss and fiber performance more than anything. Simply going with the span distance is generally not enough information for a design to be architected around. This is a way of saying, before we can give definitive answers, we need to ensure the outside plant fiber is appropriately tested to provide concrete solutions. With that stated, we can assume the 120 km span is within standard specifications. Another critical point is that optical transport systems can be architected and designed in many ways (i.e., router to router, passively filtered, amplified and ROADM). Now more than ever, it is critical that an operator considering an optical transport strategy work with a solutions integrator like WWT to ensure the proper design is implemented.
With that in mind, this question needs to be answered from two perspectives. From a DWDM filtered passively amplified solution vs. a ROADM system.
Passively amplified solution - There is some truth behind the 120 km cutoff before an optical to electrical to optical (OEO) regeneration needs to happen. An OEO regeneration occurs when a router takes in an optical interface like a 400G ZR/ZR+, runs it through the fabric and pumps it back out another 400G port. Regardless of the cutoff, some creative design techniques can assist in increasing that general 120 km cutoff. For example, we can cut the channel count. A typical 400G ZR+ amplified solution typically leverages a 64-channel filter. If we simply cut the channel count in half on some of these longer challenging spans, we could increase the effective span distance and incur more loss. Another technique we can leverage is the bandwidth can be dropped down from 400G. ZR+ optics have the unique capability of downrating the line rate bandwidth. We can incur more loss with a lower bandwidth rate while increasing the span distance. There is a little give and take with either of these strategies we leverage. If the span distance is larger than the excepted 120 km threshold, we can sacrifice some bandwidth, DWDM channels or both to achieve the desired distance.
ROADM system - The 120km cut still holds somewhat true from a hop-by-hop perspective. However, a ROADM system has the unique capability of passing through wavelengths from side to side of a node. So, it can be said that even though the hop-by-hop standard is in place, a 400G service can pass through ROADM nodes to increase the overall distance limitations.
Future considerations - Cisco is just around the corner from delivering pluggable 400G optics that will exceed the capabilities of this current generation of ZR/ZR+. These new optics will no doubt surpass present-day distance and performance limitations with increased transmit optical power and high transmit OSNR capabilities.
Mid-span amplification can be a tricky solution to pull off. It is not a standard architecture within the existing design tools available today. Personally, I would consider a mid-span amplified solution as a last resort. Still, it may be an operator's only option to deploy a 400G optical solution given the existing environment. That said, WWT has the facilities and expertise to accommodate a proof of concept. We can work with our customers to test such a solution and explore the possibilities through our advanced technology center (ATC).
800g vs. 400g largely depends on looking at the future and existing goals for the network operator. Both 800G and 400G have their place, and it simply depends on what the operator wants to leverage the most. 800G is comparatively a large, bulky and more costly system than 400G, but it has benefits. Mainly attaining double the bandwidth with better optical spectrum efficiency. Thus, most 800G solutions will allow an operator to pack the optical spectrum better and extract the most bandwidth capacity out of the outside plant fiber. However, since 800G leverages higher QAM modulation rates, the ultimate reach is very short compared to what a 400G is capable of. Now when looking at a 400G solution with pluggable QSFP-DD optics, one can benefit from lower cost, further distance, easily consumable, and better space, power, and cooling. Since 400G ZR/ZR+ optics are in a small form factor QSFP-DD pluggable, they can be plugged directly into existing routers today, making them more easily consumable across many network devices. A vast list of Cisco routers today support 400G ZR/ZR+ and is only growing more significantly over time. Moreover, since we can plug directly into routers, we can bypass transponders and muxponders, cutting costs and increasing our space power and cooling.
Overall, 800G vs. 400G largely depends on the network. If we're looking to leverage a cost-effective, lean solution that can be adopted directly into the routed layer, then 400G ZR/ZR+ will do the trick. If we have short distances and are looking to extract the most capacity out of our outside plant fiber, while cost, space, power and cooling are of no concern, then 800G is a viable solution.
In summary, Cisco Live 2022 was an incredible experience. There were terrific industry-leading professionals I was able to connect with and discuss existing problems, paths towards solutions and future network aspirations. I was able to take a couple of shifts in the WWT booth and meet some great coworkers and current/potential partners. I only leave with one heartbreaking letdown… why didn't Dave Matthews play Satellite?