Converged Optical Routing With Coherent Optics

Introduction

0:00 Hi, I'm Sterling Perrin with Heavy Reading and I'm talking today with Moran Roth from Juniper and we are discussing the future of coherent pluggable optics, which is just thinking about it recently. Coherent optics has been, had about a 14 year run, 13, 14 years commercially, but as you think about the advancements that are going on now, it's not slowing down. It's actually seems to be accelerating, which is really quite amazing for a technology that's been around so long. There's hardware, there's software. Maureen, I want to talk about all the aspects, but maybe start on the software components first. One of the big things we're hearing and seeing, and I'm sure you are too, multi-layer and multi-vendor control and management in these architectures with coherent pluggables. One of the questions that I get a lot is the management of those pluggables. Who is managing that pluggable optic in the router? Is it the optics and the optical team? Is it the router and the IP team? How does Juniper see that playing out?

Management of the transceiver

1:16 First, thank you Sterling for having me. When we talk about the management of the transceiver, we need to distinguish between configuration and monitoring. The configuration is really simple. In most use cases, the pluggable will come out in the host in the default mode with a default transmit power. And really, the only parameter that needs to be configured is the wavelengths in a DWDM configuration. really simple to do through the host, through the router CLI or the controller, the IP controller. When we talk about monitoring, the key here is supporting open APIs, because we realize that every customer, every carrier or network operator will have a different scheme of managing this integration of IP and optical. And having open APIs will allow every operator to use their own tools in order to manage that integration.

2:45 So the open APIs, Juniper is a router vendor, so you sell on the systems level, the pluggables are in there, the APIs, the responsibility for those is Juniper. That's something that Juniper has to build into its systems. Is that how it works?

3:01 Exactly.

Routing

3:02 So the transceiver is integrated into the host. The host can be a router or a switch. And the host software is the one that expose all the optical parameters through OpenConfig and NetConfiang models to a higher layer for management,

3:29 for monitoring of the other areas of interop, and it also gets into this, you know, this kind of merging of the optics and the IP layer, the routing, it really came to the top of the list in our surveys in the past, I don't know, maybe 18 months, and it's been consistently up there. This issue of plug-to-host interoperability again, I'm sure, an area that you're very familiar with in your own discussions. What do you see in this plug-to-host interoperability, the router being the host in that case, what do you see as the gaps when we look at 400 ZR plus, which is what we see as the optic that is, you know, of primary interest in telecom, as opposed to the ZR. Looking at 400 ZR plus, are there gaps in plug-to-host interop? And if so, what do you see as the main ones?

Plug and Play

4:34 So the OIF standardized a management or host to transceiver interface that is called CMEs, or Common Management Interface Specification. management interface specification. And the goal of that specification is a plug-and-play operation. It's a lofty goal that, unfortunately, today, we are still not there. We integrated and we are working on integration of multiple types of these coherent transceivers And we see differentiation in the implementation of that interface between different transceiver vendors and between DSP vendors in the market. So until we really achieve that plug and play goal, we believe that the right approach is to integrate the transceiver and the host from the same supplier. And so Juniper does that currently,

5:50 you're not a components vendor, but you have an optic that you've co-developed or I'm not even sure if that's the right term for it, but you have an optic that works with the Juniper routers today, correct?

Codevelopment

6:09 Exactly. So we, our strategy is, as you mentioned, it's the co-development of optics. We are not vertically integrated in developing all the components by ourselves. We are partnering with the building block suppliers, DSP, COSA, ITLA, the different optical subassemblies. And we have a partner that is building this module with us. And at the same time, we also work with third-party partners to integrate their transceivers into the Juniper ecosystem. So again, we are trying to foster an open ecosystem here, but we believe that there is still need in the rigorous testing of these transceiver in the host, again, until the promise of the CMEs interface plug and play is realized. Right, right, makes sense. So it's kind of a, you're open,

7:13 you will support third-party optics, but in terms of where just where the market and industry is today, you're going to kind of take a steps along that. Let me switch gears. You know, it's interesting, 400 ZR, we always say, historically, you talk about the rates going up and up whenever you get to a data rate. In this case, it is going up, it's 800 ZR, but the other really interesting trend we've seen over the past year is a downstream, I guess, iteration, 100G coherent pluggable optics, 100G or 100ZR, but I don't know that there's really a standardized term. Is Juniper interested in the 100ZR market, kind of in quotes? And if so, what do you see as the main drivers for that.

8:03 Yes, definitely. This is a very interesting development. There are different approaches to 100 gig ZR in the market. We see companies just take the 400 gig ZR plus and fix it on the 100 gig mode and then use it in that use case. We believe that the right approach, and this is the approach we are working with partners, is to develop a dedicated DSP, really, for that use case in order to lower the power consumption. And then by lowering the power consumption, this transceiver can fit into a QSFP28 form factor and not QSFPDD. This really enables it to save power and get significantly higher density in access and metro networks.

9:05 For Juniper, will those pluggables, I agree that the QSFP28 does seem to be where the market is going. It never works to kind of retrofit a higher end thing and just say it's a lower end. Do you see the 100 gig ZRs going primarily into routers or do you see lots of other hosts out there for this type of pluggable? And will Juniper do things other than, you know, kind of straight router implementations at 100 gig.

Use Cases

9:39 So we need to look at the use cases for that technology. And the main use cases that we see are two types. One is in mobile backhaul, where you need to aggregate a lot of cell site gateways into the data center, and then you need the high gig ZR capacity and reach. And the other one is in metro aggregation, where today a lot of carriers are using 10 gig DWDM. The next step from moving is from 10 gig DWDM aggregation is naturally to 100 gig, and here ZR provide the right solution.

10:26 So multiple markets, yeah, absolutely.

10:29 Yeah, last area, let me kind of look out into the future a bit, we've talked a bit about, well, we've talked about the 100 ZR trend, but for 400 gig ZR plus, today, you know, the modules that are on the market today, current iteration, do you think the performance, the specs of today's modules meet the requirements for operators? Or do you see other capabilities, whatever, you know, advancements in the performance needed to really hit mass market at 400 ZR plus?

New Application

11:14 400 GHz plus. Yeah, so, you know, with this advancement of coherent technology, starting with 400 GHz, ZR+, we talked about the 100 GHz ZR, and now we are talking about even higher performance, 800 GHz ZR, capability to take 400 GHz even longer reaches. You can say that coherent optics is taking over the world. But example of that new application is the OpenVR Plus forum just released their 3.0 specification that include a new 400 gig ZR mode that you can call 400 gig ZR++ and enable operation. If today open 400 gig ZR++ enable a few hundred kilometers reach, maybe five, six hundred kilometers, this new mode will enable more than a thousand kilometers than 1,000 kilometers of operation. So significantly improving the performance. And we believe that this advancement will enable new use cases and new capabilities for operators to use these transceivers across a broader set of applications.

12:48 Do you think, is it going to be 800 ZR+, but the operators are running it at a 400 gig rate? Is that what's going to happen, or is it actually going to be still just 400 gig ZR+, plus optic?

13:02 The key here is the power consumption, in my view. We need to see where the power consumption for the different modes is going to be. The big promise is to enable this 400 gig ZR++ to be integrated into existing 400 gig ports.

13:27 Right.

13:28 Right. And then you can use it in existing platforms, deployed in the core. If the power consumption end up to be being too high, then it will lend itself really to use in a more higher power 800 gig platforms.

13:50 Right, right. Makes sense. Great. So, yeah, we've – I think we have a new term from this discussion. We always hear software is eating the world, but maybe it's coherent optics that's going to eat the world. So there's a lot happening. Really appreciate catching up with you, Moran. Hope to see you in person in the not too distant future. Really appreciate your time today.

14:15 Thank you, Sterling. Thank you, Sterling.

14:17 Thanks for the update.