R-PHY or Remote PHY – Doesn’t Matter How You Say It.  The Hype is Real

R-PHY or Remote PHY – Doesn’t Matter How You Say It. The Hype is Real


All right, we’re live happy Friday everyone today is January 19th and this is episode 35 of get your Tech on our show on all things DOCSIS I’m Brady vault founder of the vault firm and nimble this with us is John Downey the Punxsutawney Phil of cable John is also on mute and John is also Technical expert leader at Cisco Systems John welcome back glad to have you with us again It’s not February 2nd So John you’re over a snap connection coming to us from from your new home, right? correct a little wary of the latency and stuff but We’ll see how well this does with the audio and video ok well. We’ll see how it goes Also back with us is Assaf mitat yow Harmonics the man in cable who runs it close to the speed of life as a light Asaf is also vice president of solutions in product management cable edge business of harmonica saath Welcome black backs so glad to have you back with us again Thanks for having me back looking forward to it. Thank you and Finally new with us is tal offer sci-fi fan who also recommends starting to read the Harry Potter book series Starting at book number four you skip one through three Tao is also director of product line management a terraced house So glad to have you with us today and actually tonight your time. Yeah, thanks a lot. Thanks a lot Brady excited to be here Don’t give away all my secrets as one But you know I was really glad talking to you because I’m a huge sci-fi fan myself And you gave me a tip about yeah, I really was not into the Harry Potter movies But you said the the books are really good, so I’m a huge reader And I’m looking forward to starting reading a book for Harry Potter So Fantastic lineup today, this is actually probably I think the biggest lineup We’ve had for one of our hangouts and our topic is RFI This is really a popular topic That we’ve had also in the past And I like the to kick off a question from one of our readers and the reason I want to do the question first is we all we have a Huge backlog of questions, but we never ever get to them because we always run everything up to the last hour, so One of our listeners Stewart asks. What are the benefits of moving all IP? To the fiber node with RFI other than reduced size in the headend Reduce cooling and power needs and shortening the distance does that mean CCAP architectures will be evolved to DAA So I will throw I’ll toss out the question and we’ll start with a soft Just because you’re first on my screen. It’s off. All right well first of all Saving power space and being cost-effective are really important values and the remote PHY architecture to begin with But as you get closer as you get smaller service group sizes with remote PHY architecture and a node plus zero architecture You’re actually allowing the smaller service group To reach higher bandwidth speeds so you’re actually enabling Cable operators to increase their speeds to their to their users as well as having higher performance Characteristics by removing all those amplifiers in a node plus zero architecture if you go fibre deep So lots and lots of benefits, and I’m sure I might be stealing a few of the ideas that The others will be chiming in on but you also get longer distances and more wavelengths when you go from analog to digital fiber Which is another characteristic of DAA and remote PHY? Okay Anyone else what a chime in? Yeah, I certainly agree I mean the fact that you transition from analog fibers fiber provides you with much greater efficiency and The ability to send out your signals much much more efficiently throughout again longer distances So it improves your throughput to your subscribers? other than that it provides you to bring in fiber closer to the subscriber right we’re moving the transition is to getting fiber closer to the home after up until fiber to the home so moving to remote PHY actually allows you to add the extra mile of adding digital fiber almost up to the house of the subscriber getting you closer to the final goal of of You know fiber to the home also the ability You know the fact that you still keep them Mac within the head and allows you to scale it up a little bit more right you still have better control of it inside the the the head ends and you’re able to virtualize that right you’re able to take that functionality and Virtualize it and expand it and change it and much more flexible rate Excellent All right, John John so the bottom line for me The the biggest problem or hurdle I’ve seen over the 25 years. I’ve been in industry has been the analog link being the Achilles heel the Performance is all based on that analog fiber optic link by getting rid of that analog amplitude modulated intensity modulated light and going digital The performance is much better the Emmy ours are better You’re basically taking the chipset that was in the CNCs and moving it out to the field Because it’s a digital link you can go much farther so farther distances are possible with better in the are you can run higher modulation schemes with ox-3 one so that’s On top of everything but as an RF guy. I also like the idea that The limitation of RF ports in the head then and that limitation now is pushed out to the field So when we look at service group sizing and service groups a number of service. Groups is always dictated by the number of RF ports Well, there is no RF ports anymore yet in its digital To do digital splitting combining you just go from say a 10 gig circuit into a hundred gig circuit I Started o RF you can’t do that you can’t just combine five to forty two megahertz upstream with another fight if we’re two megahertz upstream and think it’s going to be separable right I Mean a digital realm. You can do that all day long and with Moore’s law and faster speeds and cheaper prices blah blah blah We’re gonna end up with faster faster pipes, which means maybe less fiber more wavelengths, but each wavelength could be running Who knows 100 gigabit per second? Star the digital link is kind of the key point to start pushing us further towards virtualization towards cloud And giving us the opportunity to offer higher speeds, and I think that’s the bottom line right higher speeds. We can say fiber-to-the-home Yeah, we could do fiber at home with our fog But our fog is still in a long link Regard, but if we can go to a digital link then That’s the way to go Excellent I like the different perspectives here because I think that it’s unique like I here you know John kind of focusing on the physical layer aspect and you know John I think both John and I come from more the the Phi phi aspect world these both started off as RF guys Whereas thailand and the saw four are coming in from a higher layer in the in the realm there? So that’s we I think we all bring different unique opinions to it, so that’s excellent points of view So moving on Building on this I’d like to talk about the complexity of RFI so it I think that is something that we haven’t talked about in the past because we think well you know it’s It’s it’s awesome there. You know it’s it’s a nice technology that we can push out. We don’t have as John mentioned We don’t have to deal with the analog lasers So that’s one less thing to worry about But are we introducing more complex issues when we have our five devices, or is it is? It just as simple as you know we just push that RFI out there, and no there’s nothing to worry about after that So yeah, I do think it’s really complex You know to to introduce a new architecture no matter what that is right when you up a new Architecture deployments you need to make sure all the different components of that architecture works right so it’s like a new car I want to make sure everything works together. Well, so it’s always difficult to to introduce something new into the network Distributed access architecture is specifically remote PHY Brings in many more devices down to to the network to the to the places where previously there were only fast of nodes, right? So now we certain suddenly have hundreds of those active nodes out there in the field we need to be able to configure provision into a software upgrade for Monitor or make sure the health is okay So definitely you know upscales the level of Operational complexity in terms of you know the needs to manage so many active devices out there in the field Decision tools are a necessity Yeah, I mean that’s interesting you talk about software updates with a node because when anytime I worked in nodes in the past There’s just optical to electrical Converters and some wires and signal conditioning and stuff in it but software or even RFI explained more So I remote PHY basically consists of the physical structure Which is the node and the remote PHY device the RP d the sits inside the node itself Securing the remote PHY functionality basically the five functionality that was previously residing within the CMT S or C CAP Right so when you have you know the processors FPGA. That’s it inside that remote affine node They actually have soft-rock way just like you would do a C CAP software upgrade right so when you need to add more features more Capabilities, and you would want to do a software upgrades to that compactive component. That’s it inside the node So all of a sudden you have hundreds or thousands of active devices out there sitting in nodes in the street and cabinets on pole That you have to do math software upgrade for all of them You need to have the ability to Orchestrate and provision all those devices in an efficient way you can’t really count on you know CLI commands and you know type in manually different types of you know of execution command Not gonna have scale, right sure sure So so John I remember it S CTE Expo you were showing me like a small handheld device that you were plugging into and the RFI node Is that? Kind of what towel is talking about as far as like you know Provisioning these things or managing these things that is that how it’s done. I Think so I mean what we were showing at Expo is a regular smartphone with an app and that app can be downloaded today But today it works with our I know them so our I note is intelligent though It allows you to plug in through a USB from spectrum downstream spectrum Set the pad EQ. There’s no plugins had need you It’s actually controlled via software the next iteration would be you can do that over the network itself because who’s going to climb a pole with a You know a smart phone to set up the the node when I can do it from the comfort of my house So that’ll be part of the provisioning and set I foresee our P DS Being pre provisioned in the warehouse and then actually sent out with the contractors Or subcontractors to install and then we have a QR To that device, and it would do a lat/long so you could see exactly where that was deployed So you already have a pre provision for levels on the settings all pre configured The tag the contractor technician takes it out loads on the pole But maybe because of that poles in the middle of someone’s yard They have to go one pole back, but at least it will tag it once It is loaded in and installed that I know exactly the Latin long What is the GIS? Geographical information systems yeah, so I can have the location of that device and then I can keep track of it even the inode Has some security features if someone steals it it basically locks you out So there’s things that you always have to consider It was security and you know now you’re putting the network out in the field and can someone hack into it like if someone said Instead of requiring USB for me to plug in one should you Bluetooth or wireless than like well now you’re talking security? right Yep absolutely by the way the Upfront the complexity I’m saying I wish we didn’t have this complexity is we added when we went from modular cmts to distributive access architectures We had a timing server. I used to hate it the DTI server And then I thought when we went to remote PHY it was I Triple E 1588 timing Well, it’s part of a standard We wouldn’t really have that requirement, but we do have a PTP so that still requires a timing server hence it It’s much cheaper and easier to get than from one vendor back in the MCM to yesterday’s But it still adds complexity you have to set it up and make sure it’s ready to go and understand the configuration stuff like that Is the timing server and I’m digressing? Here’s a timing server still like a symmetric calm timing server, but I mean you said it wasn’t we’re not and I’m since I’m the the agnostic guy here Is it is it no longer symmetric on timing server, or is it is it just from someone else, or as a vendor dependent? I think com might have been sold to One of the companies that does provide the PTB it’s micro semi Maybe the Others before there was one vendor Talca heck Yeah, there’s a silly course as well. So there are a few of them offering different types of solutions from a server to a potential SFP+ kind of device that would fit into You know one of the chassis Zoar you know that are the CCAP core basically it does add complexity for sure right? It’s yet another device that you will need to manage and provision make sure it works make sure it synchronizes So it adds a little bit of a complexity But you know we’re thinking of more elegant solutions to try to make it less of an interfering problem cool, I like the fact that we’re not tied though to just one vendor that provides this timing sir Yeah, that is your only solution. Thank you. I will charge you whatever you charge you whatever you want Though Asaf What are your what’s your input on on this on as far as like the complexity and and some of the issues that you’re seeing? well I Think it’s an evolution. You know it’s em hav 2 for a reason and DOCSIS is sorry what what does that mean? What does that term em enter? head in architecture modular cmts was built on a technical reporting in Cable labs called MHA there wasn’t a v1 to it, but now that there’s a VGA you had a number and it’s an evolution of that and there was I would say a An inflection point of one that made it relevant is when you had silicon out there RF PGA’s out there that could do the full spectrum because once you separate the physical layer From the upper layers you want to put it out there and have it live a long time You don’t want to replace these things out in the field and now that we got a full spectrum DOCSIS into down straight 3.1 in the downstream in the upstream available is basically commercial silicon That’s something that could be separated very easily number one number two in terms of the protocol. It really is an evolution of of Modular CMT yes, we use the defi Plane and control plane there. We called it Debbie and Debbie control and over here we have Debbie you do I’d be the use for the upstream and GCP which is just a newer version of the Debbie control plane So an evolution there as well, but we’re basically taking something that was You know by all by all regards complex docks as is complex and evolving it in a way That allows us to take something that is slowly changing. That’s long-lasting in the field the node and Separating four things that hopefully are faster evolving You know the software if you have a virtualized approached in the head end and a cmp escort you can have those things Change very quickly with regards to the software upgrade It’s a great point those are active components for sure and those protocols need to be digested in the node But there’s another active component out of the field that isn’t is even deployed in a much greater magnitude Which is the cable modem and in fact? There’s some up who are doing software upgrades to their cable modems very regularly? At least weekly if not more frequently and very successfully and and certainly I would I would look at modular cmts and the and the frequency of updates to the to the Universal edge qualm right to them to that piece of it. It was very infrequently upgraded because The physical air isn’t changing very fast, so you have to have the capability to download software to these notes But they shouldn’t be changing and when they are Being to be changed is something that’s simple and straightforward and really well-defined as well as secure cable abs took Took a lot of stuffs to making sure that it’s secure and and well-defined Learning from its great experiences with the cable modems already out in the field So there’s been a lot of thought around that Re in industry, and it’s proven out in the field as well So so when you mean secure Security in the RFI like what are what are some of the things that we have to worry about from a security standpoint Well you got encryption DOCSIS has its natural encryption, but also security modems have certificates right they get issued certificates That are mapped to their MAC addresses, and so do remote by devices so these things Are out there to make sure that they’re not spooked that the link is encrypted and that that Basically there are known entities right so somebody can just put something out on a pole and expect to get serviced So that we need security and encryption and authentication Not only for cable modems, but also for remote PHY devices and again We’re leveraging known known techniques out there that are well used and applying them to this to these devices as well so does this tie back in John to what you were talking about as far as like if Someone steals an RFI device, and we know this happens. I mean people steal knows they still amplifiers they send them to other countries and Get deployed. I mean I like that concept that someone steals my RFI device It’s it’s not going to be usable so that that reduces the incentive for people to steal RFI devices and start stealing equipment is That something I mean that’s real, right Unless they steal from the pressure precious metals Yeah So I mean I like that concept And is that kind of tie back into what Assaf is just saying that it’s almost like a cable modem with a BPI plus certificate where if it’s authenticated to a Particular does it get authenticated to a cable operator or to a CCAP? How is that essentially? What’s happening? I? Mean it would have its own MAC address, or you know Identification that is then talking to the core so you have your EMT SRC cap core Cmts or. It could be a virtual Server, but that ID has to be tied back Though I don’t know all the details into that side of it, but yeah There’s it’s not like you just throw an RPD out there and think it’s going to work Yeah, I think the major you know astrophys scenario that were afraid of in this in this regard is that somebody will spoof the remote PHY note and imagine somebody pretends to be a remote PHY note It’s got access to the Mac core right to a big box that can serve up to 60,000 90,000 a hundred A hundred thousand subscribers, and they got like a full open. You know gateway to that big system. They could take it down, right Where basically with the remote PHY out there in the field we’re punching a hole in our chokehold security system and allowing people if somebody spooks a remote PHY, they can just get into the Getting to the sorry the Mac core and be able to sorry And be able to see 6000 we can do anything here. It’s automatic. I don’t even think about it You can just get into the Mac whore and be able to take down everybody else right take down ninety thousand a hundred thousand subscribers Potentially even spy on them right you could potentially take a look at this traffic that they’re getting so We really need to make sure that nobody can rip and spoof and you see I’m sorry remote right now I’m not going to be able to stop. Oh my god We’re all friends You need to be sure that nobody can really spoof a remote PHY node, so basically we have certificate we’ve got a couple of ways to secure that so there is a An operator key that is there the operator configures that and downloads that to the actual remote PHY device so that the operator is There’s also guided an identification versus the remote PHY device and we also have The Mac core is able to identify the remote PHY device that is connected to it and verify it so unless there is You know just do all handshake using standard protocols for that handshake There’s not going to be any identification for both sides right we could also somebody could also potentially spoof the The Mac core right so both sides of the equation both the macro and there are multi both needs to be authenticated and verified Such that nobody. There’s not going to be any middlemen attack in that scenario But that’s that’s really really interesting. I had never considered the security aspects of that That that would even happen I mean they go well beyond that the normal security issues that we have lots of Security is huge in the news Security’s huge everywhere, but this is just adding another layer are there other types of Concerns that RFI introduces, I mean with security is one aspect But I would imagine there’s a certain type of time that it takes for the RFI to register and do this type of handshaking what how much time does it take the RFI device to register with a CCAP relatively quickly and Yeah, I mean the protocol is there and it’s relatively quickly It’s I think the long pole in the tent and how long it takes an RPG to come up is the? Synchronizing with the 1588 as I Triple E. 1588 and depending on how that’s configured And but other than that those things come up very quickly and it happens when it gets powered up, so generally speaking It’s a pretty fast process I’d like how you don’t put a number on it Number on it fair enough we’ve experienced around three minutes For folks to wire it up on a pull and connect power when you think about it from us fan mounted experience But three minutes for a technician who put something on a strand is a pretty insignificant amount of time for them Time it takes to install it So if if it takes three minutes to come up, I mean on the first time on the power outage What happens if there are like network glitches or anything is there anything that would cause the RFI to go offline and have to re-register Yeah, essentially that could happen you know if you lose connection to the Mac core or to the aggregation router that sits on the way eventually yes the remote PHY can reboot and we kind of rear edge astir, so you’d say it’s for a cable modem the registration process is basically establishing the tunnel towards the Mac core of The GCP control that Asaf mentioned that needs to get established so that traffic can start streaming that happens very very fast as well Right, so it should be you know in almost instantaneously when you have Disconnects that the new RPD know the RPD will come back just like a new one and we registered to the Mac or What other things should operators be thinking about? like when they’re thinking of deploying our thighs I Think you started a good segue there from complexity to hurdles the IC would be Your replacing all the RF and headin out to the field so what else Replicates or produces RF that you need to now replicate on the field sweeps Dingles leakage detection signals active band set-top-box signals Return path monitoring equipment. There’s no RF and a head. End, so we’ve talked about this you Support those services, and do you even need this have this services anymore, it’s a smaller car area So that’s one aspect of it and those asked those Well are being addressed as we speak The chipset in the node the our PD has to produce the RF so video qualm DOCSIS qualms. Talks is three one qualms Out-of-band signaling could be FSK PSK whatever QPSK whatever happens to me CW carriers for leakage detection if you’re still doing that Beyond our spectrum and send it back kind of digital link so I can processed it so all those things have to be supported Almost like we have to be backwards compatible is whatever we do in the head end today Our hurdle I feel is So supporting the RF and there was another one. Oh, we’re done and see You were like well. What if it goes down? But then again you could turn back and say well what if your know goes down you don’t have redundancy there either Most people don’t right. You know they might have battery backup Tall yeah, but I’m hitting all these data subscribers. It’s still the same node It’s still the same a number of homes passed per node, so You can look at and say arm Is much better with the digital link Or you could do it separate fibers. You know you could do multiple fibers for redundant fiber but the complexity that’s when the way I think we Cisco and a lot of Industry focus on a remote fine. Not remote mech fie. I know remote max light It’s going to be part of cable labs as well Put the least complexity out in the field, but be future proof for your hardware as I see It I Can never pronounce your name as a soft like and like a sophomore, but I saw yes So as you mentioned I want to be able to have something that’s really reliable out in the field and You know the complexity is not really out in the field. It’s just the thigh, but you want to be having future proofs meaning, it’s Bring one upstream downstream that you know you can utilize for the next ten years You know it’s not going to be something where you need to replace hardware if I needed to an iOS update It’s much simpler with automation Granted you’re not just updating 10. Cm. Jess’s now. You’re updating 10,000 RP DS but you can do that in a round robin fashion much quicker and maybe you’re only updating the BGP part of it or Or an FPGA upload or something that’s very miniscule or minut versus updating entire cmts Now I agree with John I mean one of the things that that you nailed I think that is different is the end-to-end remote PHY solution that operators need to be Aware you mentioned all the right things, but it’s worth repeating auto band, right SCT e55 one or 55 to the video core Needs to everything needs to be over IP right because it’s a digital link so you don’t have any RF combining in the head end it happens a Device in Europe also. There’s FM signals HMS signals, so all those things leakage detection Those are all things that need to be there to make it end-to-end complete and PNM is an important part of this as well Brady I know that you’re There really well aware that is as well, but when you when you start looking at looking at the signals You are looking at it, because you’re receiving the RF at the head end well You’re not going to receive the RF at the head end so you have to be able To send those signals a different way over IP from the remote PHY device that you’re able to monitor and manage that centrally right and Definitely allowing your technicians to still have the capability there. They’re super capable people But you want to make sure that that their job is very similar to what it is today Yeah, they can scan a QR code And they can install that but the end of the day you want to be able to also provide The key tools that allow them to manage the technical part of their network So those are all things that make it a complete solution as well as not to mention the CMT escort so those things all exist today But of course the way, they get transmitted between the Henin and the node is not over an arc link now. It’s over IP So that that’s kind of the difference there It was one more one more Signal that we need to replicate from the RPD I forgot, but I’m glad you mentioned FM Which is you know you wouldn’t think many people carry FM, but I think in Germany they do and you know frequencies 88 to 108 on the cable plant so that has to be figured out I Think that we should just get rid of FM and do digital satellite for radio. Waye is what it is The other one was CW carries that our frequency agile maybe for existing amplifier HECS So we have to make sure that we have signals that we can produce If we still have say in those plus three node plus five Level control our automatic gain control in those amplifiers that might not be able to monitor an OFDM Signal, maybe they can’t understand a clone signal, but they still require in essence a CW you know something very stable and frequency agile and Typically when you do a CW carry the AGC is looking for 60 be higher than the digital Carrier right next to it so you have to make sure you have level flexibility as well That’s something another thing that’s been asked Yeah, I would I would take it to a different angle if you don’t mind there Yeah, I think that you know a couple of hurt. Not so much as hurdles, but more like planning items would be routing and the video architecture so on the routing side when you have to Send out the traffic from the mac core to the remote PHY node. You need to Most cases passed through aggregation routers to could be you know spine and leaf switches on the way There could be a layer 2 network Or could be a layer 3 network both would work right it really depends on the customer preferences and whatever Architecture they’ve got in their backbone But it’s important to remember this one Do the planning ahead of time to think what kind of protocols you want to deploy in your you know IP backbone right this is What carries your your signal sound oh to all of your nodes? So we need to make sure that you have the right protocol to have the right network design whether it’s with or without link redundancy as John mentions to to the RPGs themselves You need to make sure that all those switches and routers it sits on a way They support the features that you need them to support like potentially security features like Mac spec or you know the multicast distribution that you require for your video signal distribution All of those need to be thought of and make sure that you have those and also you would want to start thinking about how? Do you provision them right when you have so many? Switches and routers down there in your network and yours you know it’s fine relief you want to be able to provision them configure them The right way to make the signals flow from the Mac or to the RPD in the right way and you want to probably automate that right automate the older St. CIN their converged interface nets where interconnect Network It’s everything that sits between the Mac core in the remote PHY you want to make sure That you know what you’re putting in there. You can provision it configure it remotely You don’t need to go again manually to configure each one of those devices The second item is you need to make sure you plan your video network accordingly right a lot of customers have got different types of net of broadcast VOD SDV type of requirements in terms of narrow cast and broadcast video and you want to make sure that you can you know when the standard according to the remote PHY it Supports you know VOD SDV broadcast, it’s all supported Want to make sure that your configuration is aligned right to choose? You know replicate the broadcast to the right nodes you got multiple ad zones you need to think about You need to make sure that all your video architecture is really well aligned with what you’re planning to do with the DOCSIS and High-speed data and voice type of architecture and there are different approaches to that right there’s an approach It says okay converge everything on one mac core. There’s another approach. It’s say. Okay. Why don’t you split the core? Have a core for video record for DOCSIS That’s another acceptable approach right, but each one needs to do the analysis make sure that do the right planning by themselves Wow So I just know if I’m a cable operator I want all three of you to be right beside me when I’m making my choices to start our PI But the one thing I really like that you just mentioned tells Automation so there I mean because there’s so much complexity that goes into this, but if you can automate that How so how do you start going about automation? automating some of this Actually to start simplifying the complexity that goes into our five Deployments is that is that something that we can see on the horizon that it can it can actually make it simpler for operators? Yeah, get this stuff out there. Yeah definitely I think all of us are working on those domains so without mentioning specific products We’re all aware of the problem that you know so many different active devices down during the network basically create You know a configuration nightmare, so we need to be able to automate the perm. You know the configuration the setting up the Changing you know parameters and reconfiguration of those devices And the only way to do that is through automation right so automation is the most basic way to say okay? Let’s do a script that does CLI commands in a row right potentially But if you take it a level you know above then you would start talking about Orchestration right because you have the Mac core you have the remote PHY you have potentially the video Servers the video map you know the video course The out-of-band servers a lot of different components within the network that all of them need to work together Which brings us to a point where we actually need orchestration right we need to make sure all those devices are Working well together passing parameters to one another not you know creating any contention dimchurch configuration in the network So this is where orchestration and provisioning tools are starting to surface And you know I think all of us are kind of engaged in that front at this time Awesome and anyone else want to chime in on the automation side yeah sure I’ll just complement what what tal said I I think the industry all of us are looking at different ways to orchestrate and automate and Virtualization we covered that last time this isn’t a virtualization session, I guess But virtualization approaches out there allow you to orchestrate in a pretty effective and efficient way We’re all aware of those different tools and without side tracking or rat holing and that domain again I think having a virtualized approach enables those things to enable the operator to Remove the operator and for making too many manual decisions And making it in a more efficient way that removes some of the error errors that are associated with Such a mass configuration when you look at all the different elements that are out there So when you look at turning on a node you scan the qr-code john mentioned that earlier and something automatically? initiates Its configuration that was predetermined And allows that service to come up almost automatically because it’s lineup from a video auto band doctor’s perspective has already been done and Associated with the MAC address of that remote by node and the whole trigger there is that it’s been authenticated ahead of time They know that that node exists and is authenticated Mr.. Ation once that QR code is the seam right in the back-office? Enables it to get the right Configuration with the right yes Put the right type about AB and with the right DOCSIS Service group configuration and basically from then just letting load I’m starting to register and that’s something that exists today, and we’re all working on it As Tom mentioned and that really removes a lot of that complexity away And abstracts it away from the operator and a way to deploy these things in a really scalable way as well Awesome, thank you So if we were to look a couple years down the road two years three years Where do you guys see RFI technology going is? Is does it get smaller does it get pushed out even further into the plan? I mean do we have any visibility on this? Yeah, I’d like to see it scale up I’d like to see a steady state. I think 2018 is a is the tipping point year For the industry You know you got Cisco Eris and harmonic on the call over here I think I hope we could all agree that that remote buy solves real world problems and if you think back to se te There was a remote spy Session the day before the show started. I was lucky enough to participate with with John Chapman and Tom Coonan and and Nokia and a panel and we were expecting about a hundred 150 people and all these folks showed up And it was at least three times that So there’s a huge interest for that and I think remote spy First of all needs to be deployed in a scalable fashion in terms of how much of that Footprint is actually going to use the technology so I see that as a big thing that happens in in 2018 I believe that Jeff Payne information L. Kagan put out a report recently that showed that there’s a shift from traditional HFC nodes to to remote by style nodes and has a An estimate of how much will that will happen over the course of the next few years as well So I think the technology is really solid with full spectrum. It’ll last a really long time and Hopefully we see a lot of that inflection point somewhere during 2018 Anyone else care to hazard a guess I’ll take a risky gas Maybe not a popular one, but I think in two or three years from today. We’ll see a Great variety of solutions out there in the market. I’m sure will be widely deployed with remote PHY I’m sure integrated CCAP is still going to be Very strong potentially the number one type of solution out there in the market will see more remote smack and five potentially we’ll see more Remote Mac remote PHY sitting next to one another I think the we’re headed for a bifurcation in general And I think that in a few years from now. We’ll see a little bit of everything Remote PHY is an efficient solution for a lot of a lot of customers a lot of cases You know it’s very helpful for customers looking to again push the fiber down closer to the subscriber or It could also be very efficient in case of rural areas where you have smaller hub sites And you know more in both locations, or you could just place a node and feed an entire I don’t know what I don’t know a group of homes potentially Or you can just put a you know a remote PHY shelf which is another viable solution right it’s been set aside a small hub Where the actual Mac core is going to be sitting inside a head end the centralized head end years from now We’ll see We’ll see just about everything out there right you’ll see remote PHY in nodes in shelves Will see remote mac and files out there in many cases You know and a lot of integrated CCAP that will keep living for many many more years Okay, so you both you and John have mentioned mac v. During the conversation, and I’ve also heard other discussions of mac 5 it’s my understanding the Mac 5 we’re just we’re basically pushing the intelligence of the C CAP into the RFI I understand that correct, and you can tell me if I’m wrong but What what solution is Mac Phi’s solving over just a standard RFI? Yeah, it’s yeah, yeah, it’s a good question on I don’t feel like you know being the one the only ones talks, but yes So remote mac and phy basically takes the max functionality put that inside the node along with the Phi It tells a few things right there’s are some few benefits to that One of the benefits is that you affect you place the Mac and the Phi closer to subscriber right again? You would send out digital links of IP down to the nodes And then you will do the mac and phy Functionality and spread it down to the home on smaller service groups right to get the benefits of digital links going to that node Right with higher higher throughput the fact that the Mac and the Phi are located together inside the node allows you to provide better you know Less light latency in terms of the services because the scheduler which it sits in the Mac and the file level sits together so that potentially prevent some Latency issue that may arise from a remote PHY type of architecture This is something that a lot of discussions erasing today because of the concern about the gaming community writes about the gaming requirements for for latency within the network so that that potentially solves this problem Another you know there are simplifications that come with remote mac and phy so for example There’s the fact that the Mac sits next to the fight Reduces the need for the timing server that we just mentioned so one less item of your offshore architecture off your Ecosystem, so that’s that’s that’s nice, right? There’s no need for the actual protocol the debit protocol a TCP protocol between the headend and the node you would They will all be sitting together. You only have IP right you have IP down to the node from that on That’s that’s RF. There is no need for yet, another protocol. Which is the DEP e or remote PHY? Protocol inspect that is now emerging So some people You know look at it as a simpler more simple solution to deploy because it just put a box and there turn it on Diamond twerks doesn’t there is no dependency between the node and the mask or head and sitting inside the head end The some extend you can think of it as a simpler architecture But it’s got its own challenges like you know power and heat you know when you put the mac down there. It’s much more much more complicated to manage the heat dissipation So pros and cons to each one of them, right Yeah, and I would add to that that we’d like to see the remote mac by You know for it to get standardized right now. It’s not a standard solution. There’s a technical report that describes conceptually what what Tom mentioned? But we need a you can’t replace remote mattefies today with other remote mac by solution so the standard is it has is Having a standard solution there. I think is important so you have interoperability as well That’s a good point There’s something that’s usually overlooked and in terms of the latency that’s worth mentioning since we brought up remote mac by DOCSIS has a embedded latency with best effort services related to basically the scheduler and John Chapman actually covered this in the remote buy session at se te which is Integrated cmts as a remote by have the same latency As we are familiar today within the distances that most DOCSIS operations are deployed in which is a hundred miles or less? And the reason that is of course you have the you have to take into account that the DOCSIS Maps need to traverse Even if there is no distance. They have to cover a certain period of time Evaluated time slots And you have to wait for the modem to reply to that even if there’s no propagation delay You still have this back and forth so typically the minimum latency whether it’s remote PHY or remote mapped by you still have About five to six milliseconds best case latency no matter what? Even if the Mac is sitting one foot away from from from the modem itself And it’ll be interesting to see what happens with the standard, and I think we’re all participating and keeping well aware of that So that was one of the big Issues or perceived issues was distance and latency Our own testing to prove that with remote PHY, you know even though the Mac was potentially 2,000 kilometers away There’s 2,000 kilometers away. We simulated that with real fiber and also a latency simulator That we were able to still basically reach the line rate with DOCSIS three demands now if you’re talking about and this is upstream speed right because To request all the way down to the Mac because that’s where the Mac scheduler is and the upstream scheduling and then instead the downstream Map to say alright now. You can send on the upstream so that upstream speed is what the problem is it’s not downstream But we were able to show that with DOCSIS 3 out doing MTC mode multiple transit of a channel set And it was concatenation fragmentation that basically we can get line rate on the upstream Even with a 3 unloaded now There’s a tool Odom yeah, you might not be get better and say 3 megabits per second if it is super long distance catenation, but my feeling is If I’m going remote PHY, I’m offering higher speeds Why wouldn’t and I’d be relying on it to emot– me to begin with you know And if I had a 200 modem in the customers house Maybe that to a modem was embedded in the set-top box So that’s why it was still – a modem, but it’s per second in the upstream They only need what? 300k or whatever they did for their set-top box. You know it’s just DSG docks a set-top anyway, so The distance why is that the argument of a distance? hold a lot of water Bet the gaming and latencies II I think that one still needs to be vetted out and proven that Is there really an issue with gaming and latency with a remote PHY solution, or is there some other? Option or feature I can implement to turn around that request grant a little bit faster You could always put it on a UGS flow. I mean that’s what we do with the voice but and We also have to understand. It’s a it’s actually Impactful to the experience so far it hasn’t seemed so but obviously that that’s a that will depend on the distances We’re looking at and they’re gonna be pretty extreme distances. They’re gonna be much greater distances, and what cmts is are deployed For today we’re talking that you know multiple hundreds of miles between the the scheduler and and the remote by and the modem effectively Which is not really we can envision in the center in? Central America in central us a feeding remote PHY devices across you know entire United States So I sense I sense that there’s gonna be a lot of future debate between Mac fie and remote PHY this seems to be a interesting topic and it’s very interesting that gamers are The ones that are noticing at the most well Notice the one millisecond extra latency over a hundred miles Which is what it is right the speed of light over fibers point eight milliseconds for 100 miles And there’s speculation I still think that doxa still has the capabilities with with UGS clothes to manage gamers we could always put them on that type of Service flow, but I want to mention something that Tom mentioned, and I think it’s important is The power right we have a power budget for these things that the node that go up on the pole and and Really, it’s really hard to bring in more power into the field so Technology you mentioned how technology will evolve over the next few years probably the remote Mach 5 piece will come down and their power consumption and and of course remote by day I Like to see a bake-off but clearly I think that remote PHY is much less power consuming when you look at the power specs of of Our of our know those solutions that are available today I think that’s an important thing to mention because you can’t all of a sudden show up And you know in the head end you could always bring in more power cost more money But you have that opportunity in the field you just don’t have that opportunity right so Now that’s an important point that Tom mentioned Awesome yeah, you could expand on that actually even more you know if you want to take it another another step up the fact that you keep the Mac functionality within the head end allows you to you know expand it change it change the Devices that support the mac processing right So for example doctors treated Swan a lot of the changes the doctor spirit one introduced were done in the Mac level right if you Were to put a chip you know down the line, and the end the note, and then you want it to operate to Access Twitter to on oh now when you want to upgrade to FDX It’s gonna be pretty difficult to do you know this next technology jump to FDX for example without actually changing the device down there in the nose if you’re keeping the mac within the head and Just you know much. It’s much easier for you to actually do that to change right you’re able to change the devices Virtualize them to some extent move them you know around and because it’s all in the head and once it’s in the field The you know the the churn the fact that you need to take it out and put it back in and replace the device It’s much more expensive much more complicated, right Perfect intro to the last question for two was Final closing question. What are your thoughts on where remote PHY and will full-duplex DOCSIS? What will we see that happen will it occur in the next few years? What are your thoughts on that towel was thought worth you since you did the lead-in on it. Okay? It was an order one right? So yeah at the axis is a is a huge promise right we can’t wait to see it happen It’s going to be a very big you know relief to a lot of the upstream requirements from our customers? FDX will materialize in the remote Faiz or remote mac and Faiz both both of those and You know what’s going to happen in the next few years. It’s gonna be a little bit longer than you know I think the Advertisements would say right he still needs to manufacture the the chips the a6 for the modems you need to make sure the the That the remote files remote Mac and Phi’s will support this type of technology a lot of testing needs to get done This is an example doctor Street at Wan spec was sealed you know a couple of years ago already But still you know we’re just seeing starting to see them the vast deployments of DOCSIS 3.1 Right now all over the world, right? It’s just you know it was just starting in the last few months or so So it takes a while to adopt this new technology and really deployed that widely, but it’s going to happen for sure for 4 FDX Considering the requirements of that technology which would mean a node +0 type of architecture fiber deep type of deployment of the node, and it’s logic ok John your thoughts I mean, let’s take a step back and ask you what is FD X solving and Why do we need it? We know doctors 3-1 increased upstream the 204 megahertz which gives us more 3-1 upstream But then FD X came around and said you know how do we get the one gig on the upstream and? We’re not gonna be able to do it with Doc’s is three one across the board and Efficiently share that that speed I might be I get one gig for one customer But I need to sign up you know 300 customers share that same spectrum So how do I do that and one of the ways is get rid of the die Plex filter do a full duplex DOCSIS? introduced something called echo cancellation New technology it kind of I’d liken it back to our feed-forward days of RC core amplifiers, right? sample the signal and Delay it 180 degrees out of phase that you could cancel a signal But it’s sorted to that nature, and I don’t know if they’ll call it Fox’s 400 or DOCSIS 500 or file You know iPhones DOCSIS X or whatever? We’re going to call it, but it is going to be new hardware So the modem might need a echo cancellation the node is going to need echo cancellation the actives need echo cancellation you need smart functionality that the modems can transmit while the cmts Looks at other modems to see where the interference groups are so it has to be able to bundle that so sort of automation again So you can bundle or cluster devices into? non-interfering groups It’s called full duplex meaning simultaneous downstream and upstream, which seems really crazy when you’re an RF guy How can I transmit a 500 megahertz on the downstream or? receive 500 megahertz on downstream and transmit on the upstream 500 megahertz How can I do that well there are ways to do it, so it’s kind of interesting where those things are going But we’re trying to introduce this to solve the problem of offering one gig service on the upstream Now you mentioned about DOCSIS 3-1 Yeah, it’s being widely deployed now for downstream, but I don’t really see any upstream then no one really has a spectrum anyone That’s updated in cable plans today is going to 85 mega rows from the upstream so I might better do five to forty two megahertz for single carrier qualms a TDMA and 42 to 85 I might do a small block of OFDM a and then I can aggregate all that together and maybe offer 100 200 mega service on the upstream But until we get an application or competition to push it us on the upstream speed I Don’t know what’s going to really push us for the FDX You know something’s got to push it in it that could turn around tomorrow this new application Could come out tomorrow, then home security High Definition home security. I don’t know Even though in the cloud in the virtual side We could sit there and say and be prognosticators and say it’ll be five years before I see this well It could take someone like Amazon tomorrow to buy Altice USA and say hey, we’re gonna go to cloud and Just kind of rip everything out and go cloud Because they have that infrastructure, that’s the way they would do business Amazon and Google they read that big data service centers And not deal with the cmts business felt like that So I mean we can try to guess And and I know the hardware is not going to go anywhere. It’s soon its return on investment But it wouldn’t be hard for us to start implementing these new technologies on a as-needed basis I Have a business park that’s opening up down the road. It’s Greenfield I’m gonna do a fiber as deep as I can you asked about how deep Note plays 0 I could first see that going Fiber remote PHY taps eventually you know going down that the node is only feeding 8 customers You’re just gonna get closer and closer. It’s just the whole you know eight people sharing that Node versus one person sharing a node. That’s still cheaper so I hear we’ve come down the muddy a resounding yes from tau a Really kind of I’m not so sure from John. I saw you’re gonna be the tiebreaker okay, well I think put your money on speed is always a good bet and it’s proven out all the time and so and putting your money on HFC so FDX and the best is a bet on on cable networks, and it’s a bet on speed and and I think we’ll get there and And it’s a good bet and and In terms of I think this is funny John. I don’t think it’s 400 or 500. I think it’s an appendix in 3.1 and it’s a It’s a it’s a Sicilian right only full duplex, it’s dependent What happens when I haven’t pen to sinus? You’ll choose which appendix to remove I think that it will happen and We’re investing a lot as an industry to make sure that it does happen because it gives extra life to the investment We’ve made into the intricate Infrastructure and the cable plant them and in competing against fiber the home, and and the competition is symmetric speeds Right we still have customers that need some metric speed They need to have that upstream bandwidth to compete against that and that’s rough t Th is that you can you can tell not only a high speed downstream, but have some metric up streams And this is what will give that that lifeblood DOCSIS 3.1. By the way does already an amazing job of giving upstream Speeds if you have the spectrum if you have the spectrum and FDX Basically gives you more spectrum and the ability to simultaneously Transmit over those same frequencies, so it gives you more upstream speeds Obviously there. There is some trade-offs in the downstream and making a mid split decision or a high splits decision of where that happens But without question it’ll happen. I do agree that this first of all we’re going in record speed As an industry in terms of specifying it in terms of the folks making that the chipsets out there But then didn’t they have to be put into products and and tested and put put out there and it takes time and and we can look at the history of how long it takes for that to happen and How long it takes to scale deployments for that to happen? and but I guarantee you folks are going all of us are going as quickly as we can and the in the direction of FDX as well Yeah, I’m gonna cast my vote as a yes FDX is gonna happen because I have to recover for my lack of support for DOCSIS 3.1. And it has done so well, so I do believe FDX will happen or at least that’s what I’m thinking now So folks I’m gonna wrap up this podcast or this hangout and podcast Towe Asaf and John Thank you so much for your time today. This has been an exceptional amount of material and great topic that we’ve covered our next episode is Going to be on cable access Sdn orchestration in February I believe with John and tile if we can get tell to join us again cable access Sdn orchestration Which now you already mentioned a little bit is a tool that helps us manage all of this complexity that we’re dealing with now So we do our best to bring our audience great technical content every month if you if you’re watching us please click subscribe on YouTube or you can listen to the audio version by getting us on you I Tunes or The Play Store with your favorite podcaster, so thank you so much for watching, and thank all of our panelists for joining Again everyone have happy Friday and have a great weekend so long Alright take care. Thank you

Related Posts

Passenger Calls MIA Security Scare Over Cell Phone ‘Uneventful’

Passenger Calls MIA Security Scare Over Cell Phone ‘Uneventful’

HOSPITAL, THEIR CONDITION IS UNKNOWN AT THIS TIME. REPORTING LIVE IN OPELIKA, CBS 4 NEWS. DEVELOPING NOW AT MIAMI INTERNATIONAL

7 Replies to “R-PHY or Remote PHY – Doesn’t Matter How You Say It. The Hype is Real”

  1. Asaf Matatyaou said the R-MACPHY lack of standardization, Yes, it is.But, This situation will be improved, CableLabs has setup the R-MACPHY standardization project. So, R-MACPHY can do the IOP as well in near future.

Leave a Reply

Your email address will not be published. Required fields are marked *