Future-proof: 400G & 800G Readiness

 100G switch/router ports and transceivers are the most common and cost-effective data rates to deploy today. So how do you assess 400G & 800G equipment and when you should start investing and deploying? We’ll look at extending the life of your 100G investment, alongside what else to consider when evaluating 400G and 800G technologies in the DWDM space.

Chad Lamb: Today, we’re gonna look at some of the options that you have for transporting 100G. That 100G is the sweet spot today and comes cost per bit in your routers and switches. So that’s gonna be our main focus. We’re gonna look at the pros and cons of separating your equipment, layer 1 through layer 3 and talk about what that means. Why is it cost effective to do that?

Then we’re gonna talk about IP over DWDM, leveraging that, figuring out what the pros and cons are, and seeing if that’s an approach that is good for your organization. After that, we’ll go into the costs for transceivers – those are the most expensive components in your networking equipment. So our focus here is gonna be on transceiver costs, and then we’ll talk about the cost for transporting multiple 100G.

Then we’ll follow this up with making sure you understand what you need to do to be 800G ready in your networks if you’re not already. And that will be very helpful for you to understand as you invest for the future.

Chad Lamb: So what are your options for transporting a couple of your 100G – multiple 100G? Your first option is just have lots of fiber so you can have very short reach optics, and your routers and switches, and have one pair of fiber per 100G circuit. If you’re fiber rich, that’s a great way to do it. It’s the lowest cost in terms of transceivers, but not everybody’s got the resources to have lots of fiber.

So the second option is pretty popular if your organization is able to do this: using IP over DWDM. This is where you install the DWDM transceivers right in your switch or your router. You have to have some understanding of the optics. You have to buy mux/demux filters, amplifiers, etc. The DWDM transceivers are expensive, much more so than the short reach stuff that’s in option one. But it is what you have to do if you want to try to put more 100G circuits on a single fiber pair. And then, of course, there’s the optical engineering expertise that you need in order to be able to do this.

And then lastly, the more traditional approach would be to use transponders and muxponders. There are more boxes to manage when you have to do that, when you go that route, so that would be a drawback for going with transponders and muxponders. But the transceivers are still the same cost as in option two, and then arguably it’s a higher capex, but a lower opex than IP over DWDM. So that, that really depends. It’s subjective on the types of transponders you buy and what your organization’s capabilities are in terms of being able to support IP over DWDM.

Chad Lamb: So why would you want to consider separating the layers? For DWDM, the critical component is the transceivers, they have a life cycle of at least 7 years – typically 10 years. So that means the usefulness of that component is gonna last a long time. Whereas if you look at a switch, those are basically printed, right? They’re on a dye that they can lower the line-width and in about 1.5-3 years, you’re gonna get a new generation of a switch chip.

And then for a custom ASIC, that’s the requirement for a router, you’ve got a longer lifecycle for the chip, because that’s a full custom chip. But you can expect that a router with that inside will be about a 3-5 year life cycle. So that doesn’t mean that you’re gonna get rid of these things in 7 years, 1.5 years or 3 years, it’s just the next generation of that thing will be ready in that time.

Chad Lamb: So for example, let’s say you buy a transponder – that’s a layer 1 product. That also includes switching capabilities – a layer 2 feature. That transponder should be deployed for at least 7 years because of the cost of the transceivers, and you’re gonna have to do that to recoup your investment.

But in that same timeframe, there’ll be about 3 generations of switching technology improvement. Not a big deal if you’re not worried about looking to capture the latest generation of switch chips, but there is a cost-improvement and power-draw improvements. There’s a lot of reasons that you want to go to the next generation switches, but you’re not going to be able to because you combined layer 1 and layer 2.

Chad Lamb: Let’s just talk briefly about the pros and cons of IP over DWDM. This is a pretty popular approach that people go with for trying to improve their capacity or extend the reach of their network. So some of the pros: you don’t need to buy muxponder/transponder equipment.

I think that’s the main draw. Why buy an extra box if you don’t need to do that? So therefore there are few boxes in your network and fewer boxes to manage, fewer boxes to get certified on – lots of advantages there. And it’s less rack space, power, cooling. There’s less stuff.The capex is going to be less when you consider that you don’t need to buy the transponder. You do need to buy some of the other capabilities that DWDM networks require. But generally speaking, it would be a little bit lower in capex.

DWDM capable switches and routers will be needed. So what does that mean? You have to be able to tune the lasers. You have to be able to get the information out of the lasers from the operating system. And a lot of switches and routers are not capable of doing that, so you have to be careful on what you deploy for switches and routers.

What are the cons? Well, it’s higher opex, so your staff is going to need to be able to figure out how to do this deployment. They’re gonna have to do a little bit of optical engineering, they’re going to have to figure out what kind of parts to buy. There’s a lot more opex overall to keep track of the wavelengths and everything else that might be going on in your network. And that segues into scaling difficulties. So as you add more capacity to an IP over DWDM solution, it gets harder and harder to scale.

Management of layer one equipment, transceivers in particular, is a weak point in many of the switch and router software packages that are out there. They’re not used to all of the telemetry and the information that’s available in the DWDM lasers. And so you’re left with an inadequate amount of information to monitor your network.

And there’s no demarcation point. So this can be important for a lot of different industries where you need a hard line in the sand between the outside point and the inside point, that would be your demarc point. And then there are difficulties if you’re trying to integrate other aspects of layer 1, other features like, BERT, switching protection, and OTDR features. These are not easily integrated with an IP over DWDM solution.

Chad Lamb: So let’s talk a little bit about cost. The DWDM transceivers are the biggest cost component in your network. So this is an important item to look at, and it’s pretty straightforward to compare these components. So if you’re just doing non DWDM networking, a 100G LR 10 kilometer reach is about $2.50 per gigabit. Iif you need to go further than 10 kilometers, you can do the ZR4. That’s about $25 per gigabit. So these are obviously your first choice. You want to be able to reduce your costs, but this is not DWDM. You can’t go any further or expand any more on your 100G circuits with those transceivers.

So next you get into the DWDM transceivers. So we’re looking at the coherent devices. The next one on the list there is the 100G coherent. And these are all 2025 prices, so these are relative numbers.

Your mileage will vary depending on your sources for your transceivers, vendor A versus vendor B. Are you buying it from a non-preferred country with high tariffs or are you importing it – or not importing it at all and buying it directly in the US? So these numbers are really relative to your organization. But a 100G coherent is gonna be about $5,000. So that price drop jumps up to $50 per gigabit – that’s starting to get up there. But then if you get to the 400G stuff, the 4x100G coherent ZR+, and the brights, you can see how the cost per bit is really dropping significantly. And when you get to the 800G, which will be readily available this year, now the price is really starting to get down there where it makes that a very attractive solution.

So how can you take advantage of this pricing? In the top diagram, that’s an IP over DWDM example. So in this example, we’re transmitting 4x100G using the 100G coherent transceivers. So you can see in this example, the cost of just the transceivers is about $100 per gigabit. Plus you’ve got your equipment, of course, the mux/demux filters, but just for the transceivers. If you go to a transponder-type solution, then the price drops significantly, because you’re able to take advantage of the fact that transponders can, can use the 400G optics and you can aggregate multiple 100G that way.

So it’s really something to think about as you grow and expand that this capability is not gonna happen unless you buy a 400G router board or 800G router board. Those are very expensive. Also note that if you did go with the IP over DWDM solution here. You still need to buy DWDM-capable switches and routers for that approach.

But I think this clearly shows the cost per gigabit is significant if you go with a transponder-type solution.

Chad Lamb: Now, let’s just talk about what it would take to make sure you’re 800G ready. So if you were to go with either IP over DWDM or the transponder/muxponder solution, you want to make sure you’re buying and putting in the infrastructure you need to be 800G-ready. So the main components that you have to be aware of are the filters that are going into your network, the amplifiers, and if you have them, dispersion compensation modules.

We’ll talk briefly about that, but dispersion compensation is not needed for coherent transmission. So if you have those in your network you really wanna get them out to be able to take advantage of 400G and 800G.

On the filter requirements, it comes down to the bond rate of the technology – the rate the data is actually moving. And this just summarizes it. I’m not gonna go into the details here on what this all means, but suffice it to say that the right column there is telling you what size filters you better have in your network if you want to be 100G ready, 400G ready, or 800G ready.

And this is showing an example of if you were to use what’s called a flex grid filters. These are programmable WSSs or ROADMs. You can really jam more data into your fiber and it’s much more spectrally efficient, a much more efficient use of your fiber to try to transport your signals down. So in the top example, that’s using 75G channels.Those are a hundred gig ready channels.

And the next example, if you were to try to be 400G or 800G ready you have to have wider channels. And so you have fewer that you can get down that fiber. But with a flex grid, you can mix and match your data rates and be much more efficient in your use of the fiber. And those are the basic considerations.