Cisco Optics Podcast Ep 56. The smartest data center operators use multi-mode fiber - do you? (5/7)

[00:00:09] Hello everyone and welcome back to the Cisco Optics Podcast where we talk about Plugable Optics for Networks. Multimode Fiber has been the workhorse for fiber optic communication within a data center for years, as well as a choice for Ethernet backbones and office buildings.

[00:00:23] It has a rich history that goes all the way back to the first optical fiber standards when 100 megabits per second was a huge data raid. Anybody remember those days? This is episode 56 and we continue our conversation with Hao Dong, Market Technology Development Manager for Corning's Optical Communications business.

[00:00:43] We get into the various Multimode Fiber Standards, Transceiver Wavelengths, and Future Transceivers for Multimode Fiber. Hao Dong is a Market Technology Development Manager for Corning's Optical Communications business. Within this role, he focuses on technology and market trends to identify,

[00:00:59] evaluate, and develop leading optical network solutions that formed the backbone of today's connected world. Hao has over 18 years of industry experience and has held various positions in engineering, development, and marketing within the fields of optical components, fiber lasers,

[00:01:14] and optical communications. Hao holds a Bachelor of Science in Physics from Wuhan University and a PhD in physics from the University of Connecticut. One quick note before we start, you'll hear Hao mention the need for a 100 meter reach fiber. Since we recorded this,

[00:01:28] Corning has actually released such a fiber, it's their ClearCurve OM4XT Multimode Fiber. And now join me as I talk with Hao Dong. OM3 and OM4 as a laser optimized Multimode Fiber that was considered as a workhorse for the new installation nowadays.

[00:01:56] So what did they optimize? So OM2 is for LEDs, OM3, what was the optimization? Like what's different about it that makes it optimized for lasers? Well, as I mentioned, nowadays for 50 micron multimode fiber, they are grid indexed fiber.

[00:02:18] So the reason behind that is with the grid indexed fiber, it will try to mitigate the model dispersion. So higher order mode at the edge, seeing less refractive index and fundamental mode or lower order mode in the center, they follow a shorter path,

[00:02:45] but they see a higher refractive index so that the speed is slower. So with this arrangement, eventually it kind of like a balance the time difference. So that's what's going to happen in the ideal world, in the perfect world. However, in practice it's very difficult to

[00:03:06] control the profile precisely to match your need. So that's, in terms of the control, there is a control of the profile, there is a difference between OM2 and OM3 and OM4. Eventually, it gave you different information carrying capability or the capability to address

[00:03:29] those model dispersion issues. So when you say that controlling, talking about even on paper, it's difficult to come up with a grid index design that perfectly compensates the longer path or the shorter path but a higher index versus the higher index but the longer path.

[00:03:54] Yes, on paper, the standard defined the different values. For example, for you must have heard the OM3, the ENB values as OM3, that's 2,000 MHz kilometer. For OM4, that's a 4,700 MHz kilometer. So these are the values for the ENB or the so-called information carrying capability.

[00:04:20] But when you think about how to link those values with the profile, it's very difficult to quantify the profile details. But that's giving you an idea how the information carrying capability and eventually with the different fiber, how it will impact the optical transmission performance.

[00:04:53] So what about OM5? Is there anything you want to say about OM5? Yeah, sure. So OM5 is a relatively new type of multimodal fiber. That was basically the OM5 was introduced to support short wavelength derivative multiplexing or SWDM. I think it was introduced

[00:05:20] by TIA standards in 2017. But essentially, OM5 could be considered as a subcategory of OM4 fiber because if you look at the ENB values for both OM4 and OM5, they share the same effect on model

[00:05:40] bandwidth or ENB values at 850, which is 4,700 MHz kilometer. The only difference between OM4 and OM5 is that for OM5, they also specify the ENB values at 953 nanometer because like I said, OM5 was introduced for SWDM. So for SWDM, you will need the wavelengths ranging from 850

[00:06:13] all the way to 940. And that's why you need to define, you need to specify ENB values at higher wavelengths. And for people who aren't familiar, SWDM is the transceiver type where four different wavelengths are multiplexed onto a single multimodal fiber.

[00:06:34] Right? Exactly. Yeah, typically I think the industry considered 850, 880 nanometer, 910 nanometer and 940 nanometer. So all of the four wavelengths will be through WDM, you combine them together and they're sending the signal through the one single fiber strength.

[00:06:57] The benefit of course is very obvious, right? You could save the use of fiber. But the problem is that I think there are two problems. One is that the cost is much, much higher, right?

[00:07:11] Compared with the parallel or even by dye. And the second thing is it doesn't support a breakout. So sorry, just to review, to recap. So it sounds like OM4 was specced only at 850, you said, nanometers? Yes, OM4 is only specified at 850 nanometer.

[00:07:40] And so that predated SWDM4 I'm guessing because at the time of OM4 there was no need to specify at any other wavelength. That's right, that's right. Because most of the multimodes before

[00:07:53] SWDM, most of the multimodes will be based on 850 LA operation. So even though besides 850 LA and SWDM, there is another multimode application which is called a by dye. But the by dye typically is based on 850 nanometer and a 915 nanometer. 915, not 950.

[00:08:21] 915, yes, 850 and 915. From that standpoint, you know, you also, you don't really, technically you don't need the OM5, right? OM5 sounds like overkill for by dye. Because OM5 goes all the way out to 950, 950, whereas by dye only goes out to 915.

[00:08:48] That's right, yeah. What are your thoughts on future proofing when it comes to multimode fiber? Because we haven't really talked about the future that much. We've talked about history quite a bit. But what do you see in the future and what thoughts come up when you

[00:09:07] hear future proofing? Well, to, I think to answer the question, we probably could look at the trends. What is driving the trends in data center? So obviously there is a bigger driver for higher

[00:09:26] speed. Today if you look at the switch, 25.T switch and even the 51 terabit per second switch capacity is being deployed. So for higher switch, obviously it will be driving the higher speed of transceiver eventually. So for single mode aside, 100G per LAN based

[00:09:53] client transceiver, I think that's a new norm. If you think about the 400G transceiver, there is a 400G DR4, FR4, LF4. Those transceivers are based on 100G per LAN single mode already. So obviously for short reach application for multimode, we need something to be compatible

[00:10:14] with that. So in our opinion, I think it's critical to have a 100G per LAN based on multimode available. And the distance wise, it should be able to cover distance up to 100 meter. Because that's what people are accustomed to deploying at the lower speeds,

[00:10:36] up to 100 meters in their data centers using multimode fiber, right? Exactly. So that means the two things, two specific points. One is that from the optical standpoint, it need to be 100G per LAN based and from the rich standpoint, it need to cover

[00:10:54] 100 meter, especially for OM4. So speaking of the future proofing, I think that's what the future should look like for multimode. Again, 100G per LAN based with the distance up to 100 meter. Yeah, if you look at what the standards or MSA are doing, basically they are following this

[00:11:18] path. For example, IEEE defines a 850 nanometer based application for 100G per LAN, IEEE AO2.3DB. I think this standards start about two years ago and by September 2022, the standard association is already approved that. So now 100G per LAN... Did you say 100G or 400G per LAN?

[00:11:45] Well, it's 100G per LAN based but the data rate for the transceiver will be, I think, all the way up to 400G. So 4 by 100G basically an SR4. That's right. Yeah. SR4 meaning there's four fiber pairs. So each fiber pair is a duplex, 100G.

[00:12:18] Yes. So in total you will need the A-fibers for the module. So that'll be, yeah, A-fiber based solution for 400G data rate. And this is something that OM4 can support? Well, so that's a good question. So IEEE defined OM3 for 60m and OM3 for 60m, OM4 for 100m.

[00:12:50] So like I said, for future proofing solution it need to be supporting the distance up to 100m. But if you look at the history, the industry basically follow kind of like a golden row,

[00:13:04] 70m OM3 and 100m OM4. For example, at 400G that's a case for 100G SR4 it follows 70m OM3, 100m OM4 and 400G, even 400 by that it follows the same arrangement. But the problem is that for this newly defined IEEE standards, the OM3.

[00:13:42] That was the fifth part of my conversation with Hao Dong. Next time we'll get into future high speed standards. We have a new website. It's optics.podcastpage.io. You can either listen there or use the same podcast platform you've been using all along. Please subscribe. Better yet,

[00:14:01] leave a review, especially if you've been using Apple Podcast. Remember, we're part of the Cisco Podcast Network where you can find other great Cisco podcasts too. We also have educational videos on YouTube. Just go to youtube.com and search on Cisco Optics. Thank you for listening.

[00:14:16] This is Pat Chao, product manager at Cisco Optics. The next episode is part five of my conversation with Hao Dong. Until next time.