Sponsored Links

50Gbps service

RG13

Pro Member
As a follow-on to the article, I thought it would be good to add a discussion here given the comments format is a bit of a PITA.

Figured I'd start off with my 2p.

ToS, ToS, ToS.

I have the 8Gb package because why not. There has been, up to this point, absolutely nothing that would, in a single use scenario, leverage that throughput. The closest is Steam, pulling 2.2Gb at best, which doesn't use P2P for distribution anyway. Any P2P client notwithstanding, its real advantage is heavy, multiple stream use. Even in a household with several people streaming 4K content concurrently, you'd never hit that. Even multiple Kaleidescape streams won't come close.

While I've not read the terms of service in depth, I'd imagine that use cases where you could properly leverage this throughput will be frowned upon. Hosting a website with little traffic is one thing. Hosting a file server with multiple clients may be something else.

That said, while a 50 (40)Gb package is cool on paper, as alluded to in the comments, virtually no one will be able to leverage it.

As for the suggestion about overall capacity, offering a single client this doesn't matter. Now, I'm not 100% familiar with OLTs and their internal topology, but the suggestion with respect to split ratio...does it matter? The main downside is knocking out that many more subscribers should a hardware issue present itself. If over-subscription is a concern, would you not simply rate limit throughput for a given subscriber? Now, I hate the notion of intentionally creating an over-subscription topology. Since the specs of the OLT intending to be used are unknown, I'll use the 6330-48 as an example.

480Gb at best throughput. It claims it can handle 6,144 subscribers. Probably goes without saying you're not giving each 10Gb (or maybe it does?). Even at 1Gb you're already nearly 13x oversubscribed. I appreciate the case of averaging out data patterns and the like, but still.

Correct me if I'm well off base here with respect to how OLTs are generally over-provisioned.

Only model ref I've seen for the 6400 series is the 6405-4. 4-port 50G interfaces. 2x uplinks of I'd assume 200Gb. Overall less capacity than my example.


Adtran Q: "What difference will 50GPON service make to your customers?"

Jeremy A: "...I'm not sure."

Truer words have never been spoken.
 
  • Like
Reactions: bon
It's really only going to be any good for provisioning multiple 10/25 customers, honestly. And right now, how many of those do they have on a single OLT? I bet not many! I know there are a handful but this really is willy waving with the lack of any serious application. However, I think it is important that companies are out there pushing the envelope and deploying this equipment. I'm sick and tired of the backwards attitude a lot of people have to internet connectivity in this country. A lot of people will complain this is pointless, and they're probably the same people who'd be content with their 40Mbps FTTC connectivity.
 
As per my comment on the article, I don’t think this technology choice is directly aimed at an end user; it will be for capacity density reasons. They may offer higher speed packages but I suspect it will allow them to increase the split ratios and reduce capex and opex (less equipment and less power consumption). They will probably offer increased speeds too but if they do, I highly doubt it will be anything more than 20gig.
 
Sponsored Links
However, I think it is important that companies are out there pushing the envelope and deploying this equipment.
I think the net benefit is the overall capacity of the network as a whole. I believe, as mentioned in previous articles, their back end has been upgraded and is continuing to be upgraded to support it. It's a win win for the consumer at the end of the day, removing the ridiculous notion of asymmetric speeds and increasing that ability to sustain what's actually being provided.

I suspect it will allow them to increase the split ratios and reduce capex and opex (less equipment and less power consumption). They will probably offer increased speeds too but if they do, I highly doubt it will be anything more than 20gig.

Given what we can glean into the OLT to be used, not knowing of course what its cost is compared to what's already been deployed, I'd expect both to go up a bit as the density/capacity for the 50GPON unit is less. I'm looking at it strictly from an overall throughput perspective of course.
 
I think the net benefit is the overall capacity of the network as a whole. I believe, as mentioned in previous articles, their back end has been upgraded and is continuing to be upgraded to support it. It's a win win for the consumer at the end of the day, removing the ridiculous notion of asymmetric speeds and increasing that ability to sustain what's actually being provided.
Absolutely right.

However, one thing to bear in mind of course, this only covers the PON side of it. The connectivity is only as good as the uplinks, feeding the OLT mind you.
 
As I see it, deployments and upgrades like this are not about the "package speed" to individual subscribers but about the elasticity across the network as a whole. Deploying this kind of PON technology also somewhat forces them to think about backhaul capacity too.

Over-provisioning PONs is mostly fine because most people are light internet users (even those that think they're heavy users aren't often that heavy either) and therefore there's a cost saving to be made compared to active ethernet per subscriber.

However, sometimes network utilisation will shoot up due to external factors — some examples might be a new PlayStation game becoming available, a large iOS update gets rolled out, some sporting event happens that results in an uptick in streaming or whatever. It's like when people rush to put their kettles on during the ad breaks and put a sudden strain on the energy grid.

It is good to be able to meet that sudden and unexpected demand whilst being able to continue giving everyone else a reasonable service level.
 
I think there isnt much of a point of a individual user fully utilising such a connection, not only is it a case of looking for a use case that would benefit from it, but also the equipment required to utilise such throughput.

However with that said, a ISP pushing the technology is never a bad thing in my view, and I rather see this than an ISP being content with keeping infrastructure on GPON.

I think the two obvious benefits that stand out are marketing benefits, and increasing that shared PON capacity. Whether Netomnia actually have any noticeable contention on their XGS-PON I guess only they know, but I can see the marketing benefits, this will draw eyeballs to the brand and its surprising how the human mind works, in the PC world as an example the company with the best flagship product can gain lots of sales on mainstream products, whether the same can play out in the broadband world no idea, but I suspect since the bulk of FTTP cost is in the rolling out of the cables, this might be worth a roll of the dice for the company, it may also possibly have a business customer plan linked to it.

Bear in mind as well Netomnia still seem to be struggling to get partners (surprising for me), they may need to offer something unique that other wholesalers dont have to draw partners in.
 
Sponsored Links
As per my comment on the article, I don’t think this technology choice is directly aimed at an end user; it will be for capacity density reasons. They may offer higher speed packages but I suspect it will allow them to increase the split ratios and reduce capex and opex (less equipment and less power consumption). They will probably offer increased speeds too but if they do, I highly doubt it will be anything more than 20gig.
Increasing split ratio unlikely. Wide deployment any time soon, as in lighting a PON with it before an order requiring it is received, unlikely.

Any saving that would be made from doing so would be more than offset by the much higher cost of the 50GPON port, the much lower port density of the chassis and the much higher cost of the ONT relative to XGSPON. A 6320-16 will give up to 2048 premises passed in 1.5 RU.
 
As for the suggestion about overall capacity, offering a single client this doesn't matter. Now, I'm not 100% familiar with OLTs and their internal topology, but the suggestion with respect to split ratio...does it matter?

Yes. Every extra split introduces another 3 dB of loss on the optical path. The ISP would also want an extra split at the headend near the OLT for the co-existence element to take a new PON standard out into the PON.

On the matter of capacity I can't discuss too much but think you'd be astonished at just how many customers, even with the odd one on 8 Gbit, that'll fit with no visible contention into 20 or 30 Gbit of backhaul let alone the pair of 100 Gbit uplinks to the exchange's switches/routers each OLT has. With most customers purchasing a gigabit service using the old school measurements contention ratios well over 100:1 are pretty comfortable right now 99.95% of the time as long as the highest single customer link isn't too close to the lowest backhaul/uplink capacity..
 
Yes. Every extra split introduces another 3 dB of loss on the optical path. The ISP would also want an extra split at the headend near the OLT for the co-existence element to take a new PON standard out into the PON.

On the matter of capacity I can't discuss too much but think you'd be astonished at just how many customers, even with the odd one on 8 Gbit, that'll fit with no visible contention into 20 or 30 Gbit of backhaul let alone the pair of 100 Gbit uplinks to the exchange's switches/routers each OLT has. With most customers purchasing a gigabit service using the old school measurements contention ratios well over 100:1 are pretty comfortable right now 99.95% of the time as long as the highest single customer link isn't too close to the lowest backhaul/uplink capacity..
I think this is an important point - because the speed jump is massive between 40Mbps and 1000Mbps, the amount of time connections are being utilised is massively reduced.

Let's say we are downloading a 5GB file -

@40Mbps = 1049 seconds
@1Gbps = 41 seconds

So you can download 25 x 5GB in that same period if you are the 1Gbps user - it means less people are using their connections at the same time. I think this makes the old contention ratios a bit out of date now.
 
As for the suggestion about overall capacity, offering a single client this doesn't matter. Now, I'm not 100% familiar with OLTs and their internal topology, but the suggestion with respect to split ratio...does it matter? The main downside is knocking out that many more subscribers should a hardware issue present itself. If over-subscription is a concern, would you not simply rate limit throughput for a given subscriber? Now, I hate the notion of intentionally creating an over-subscription topology. Since the specs of the OLT intending to be used are unknown, I'll use the 6330-48 as an example.

480Gb at best throughput. It claims it can handle 6,144 subscribers. Probably goes without saying you're not giving each 10Gb (or maybe it does?). Even at 1Gb you're already nearly 13x oversubscribed. I appreciate the case of averaging out data patterns and the like, but still.

Correct me if I'm well off base here with respect to how OLTs are generally over-provisioned.

Okay I see. This is a non-issue. Statistical contention makes it far less likely that 6,144 subscribers are generating 480 Gb/s of traffic than that 128 customers are generating the 10 Gb, we'll ignore FEC, that would saturate the PON port. As links get bigger customer usage smoothes out better and better as it's just less likely folks are simultaneously consuming capacity as usage even when streaming most of the time is bursty so oversubscription ratio can go higher and higher.

Based on last time I saw the numbers for an ISP network 6000 subscribers would happily fit in 1/10th of the 480 Gb you mentioned for now. This will change in a few years as all TV goes IPTV but for the foreseeable fitting 128 punters onto an XGSPON port is more problematic than fitting 48 XGSPON ports onto 2 x 100 Gbit/s of uplink.
 
Sponsored Links
Increasing split ratio unlikely. Wide deployment any time soon, as in lighting a PON with it before an order requiring it is received, unlikely.

Any saving that would be made from doing so would be more than offset by the much higher cost of the 50GPON port, the much lower port density of the chassis and the much higher cost of the ONT relative to XGSPON. A 6320-16 will give up to 2048 premises passed in 1.5 RU.
So would you say it is a largely pointless exercise? I can't really see the benefit to an end user, other than willy waving rights by claiming the top spot amongst retail ISPs.
 
So would you say it is a largely pointless exercise? I can't really see the benefit to an end user, other than willy waving rights by claiming the top spot amongst retail ISPs.
Nope.

Gives Netomnia an opportunity to try out the technology, gain experience with it, give feedback to Adtran, shape the product more towards their needs, and to both sell a 'clean' 10 Gb service and a 40 Gb 'halo' service if they wish. If they go down that path and the customer places the order they install the OLT in the exchange, the ONT in the customer premises and provision it on a 36 month business contract same as the ones on the original 8000 Mbit service before last year's residential one.

The headlines and marketing are always good. Innovation is always good.

The technology is bleeding edge. Adtran do not have hundreds of them in warehouses waiting to be shipped out to customers to be installed for mass market use. The ONTs are not practical for shipping to some random that fancies 2 Gbit/s for his PC's house. They are rack mounted just as all the Nokia 25GPON ONTs were until earlier this year.

Field trials then the release. As the technology is deployed more widely prices go down, ONTs eventually get to a more friendly price and form factor and they can integrate 50GPON more widely as required to relieve capacity on XGSPON and to satisfy customers' need for speed.

A lot of stuff happens that doesn't have an obvious, immediate benefit to all end users but is actually really good stuff in the medium and longer term. Rollouts have to start somewhere, products don't spring into life fully feature complete, functional with massive supply ready to go at mass market price points.

It is really cool technology, too.
 
Nope.

Gives Netomnia an opportunity to try out the technology, gain experience with it, give feedback to Adtran, shape the product more towards their needs, and to both sell a 'clean' 10 Gb service and a 40 Gb 'halo' service if they wish. If they go down that path and the customer places the order they install the OLT in the exchange, the ONT in the customer premises and provision it on a 36 month business contract same as the ones on the original 8000 Mbit service before last year's residential one.

The headlines and marketing are always good. Innovation is always good.

The technology is bleeding edge. Adtran do not have hundreds of them in warehouses waiting to be shipped out to customers to be installed for mass market use. The ONTs are not practical for shipping to some random that fancies 2 Gbit/s for his PC's house. They are rack mounted just as all the Nokia 25GPON ONTs were until earlier this year.

Field trials then the release. As the technology is deployed more widely prices go down, ONTs eventually get to a more friendly price and form factor and they can integrate 50GPON more widely as required to relieve capacity on XGSPON and to satisfy customers' need for speed.

A lot of stuff happens that doesn't have an obvious, immediate benefit to all end users but is actually really good stuff in the medium and longer term. Rollouts have to start somewhere, products don't spring into life fully feature complete, functional with massive supply ready to go at mass market price points.

It is really cool technology, too.
Can echo a lot of this given I was involved with the first commercial deployment of 25GPON with Nokia - As XGS points out, we were able to feedback data to Nokia that they didn't necessarily have, and we proved the point that the real benefit for right now would be a symmetric 10Gbps product. This is something that could be a contender against EAD - which Openreach make considerable amounts of money on.

Maybe everyone sees the "OMG 50Gbps" headline and forgets about all the other potential use cases. :)
 
I think Netomnia have probably correctly decided that being a launch partner on stuff like this, getting kit priced accordingly and then announcing it to the industry creates a halo product and a buzz that would cost an awful lot more in ad buys. By the time it's filtered through a couple of non-technical sources the headline is "YouFibre will do 50Gb broadband and the competition can only do 1" and the job is done. It's also a great way to get people who are somehow adamant they need it to pay a disproportionate amount towards the physical network build costs.

Being the first customer onto their 50G-PON service where the only ONT option is a rackmount box would presumably add a non-trivial amount to a power bill - though considering 8Gb is £100/month we can assume a product built on this to be £200+ and maybe the energy cost becomes insignificant to someone with that budget.
 
Yes. Every extra split introduces another 3 dB of loss on the optical path. The ISP would also want an extra split at the headend near the OLT for the co-existence element to take a new PON standard out into the PON.

On the matter of capacity I can't discuss too much but think you'd be astonished at just how many customers, even with the odd one on 8 Gbit, that'll fit with no visible contention into 20 or 30 Gbit of backhaul let alone the pair of 100 Gbit uplinks to the exchange's switches/routers each OLT has. With most customers purchasing a gigabit service using the old school measurements contention ratios well over 100:1 are pretty comfortable right now 99.95% of the time as long as the highest single customer link isn't too close to the lowest backhaul/uplink capacity..
Taking the backhaul and further connectivity back to the core out of the equation as that's about as straightforward as it can be, port capacity on the client facing side I suppose falls into the notion of assumed traffic patterns. It's unlikely that all clients behind the split will be loading up their lines 100% of the time...so I'd gather it's a bit of best guess. (EDIT: I'd typed this up before your other reply).

As for the insertion loss, I assume if the optics power budget is high enough, that determines how many you can get. For example,


...and let's call it 11dBm. My service is currently at -18dBm. We could assume that this optic would support up to maybe nine clients off that port (not taking other variables into account of course). I know for a fact that I'm the only one down the street that's on the 8Gb package. I also know that it's barely touched in terms of throughput. At least two other houses are on YF, of which I believe are on 1Gb packages. Where the OLT is and its connection in relation to our street I can't say.

I don't know what kind of split ratio as compared to DSL and a corresponding DSLAM for those clients would be, but I would assume similar issues apply, not taking into account the asymmetric nature of DSL or cable for that matter. Perhaps the inherently faster speeds mean the corresponding lines are utilized less overall, thus the risk of real contention is lessened?
 
Sponsored Links
I guess a lot will depend on how much the ONTs are, and how they're swapped. If they're £20 each and they post them out for the end user to swap out, maybe they can move the bulk over and gain the elasticity benefits.

If it's a case of them being more, and needing an engineer, I don't see a massive swap out. Maybe the 8G package will become 10G delivered over 50G.
 
It's more having to pay more for ONTs with more powerful transceivers alongside having to pay more for the OLT transceivers that's the thing with the split ratios. The Adtran transceivers don't do +11 dB transmit power either, off top of head they max at +6.

The Adtran kit stops at 1:128 split ratio, which is plenty.

The ONTs for 50GPON will be a lot more than £20 for quite some time. The XGSPON ONTs remain a fair bit more expensive to this day.
 
Would also be interesting to know for a typical PON what's the mix of speeds and actual customer numbers.
 
I think this is a great thing to have in the market and it will certainly have a product halo effect for Netomnia and the ISP's who choose to use their network. Even if a customer doesn't want this package it's nice to know you're with a service provider who is innovating and can deliver more speed than you ever think you'll need.

For me on a personal level, there was one thing I'd like them to do with regards to the service, this isn't really a big deal or anything just something I'd like done.

I mentioned to XGS_Is_On on Discord yesterday that with this kind of speed the terms of service with regards to the fair use policy (FUP) needs to be more concrete than it is usually which is to say, wishy-washy with no numbers and asterisks next to the word "unlimited".

Like even if it had a FUP which says, you can do 100TB a month and after that we reserve the right to throttle you until your next billing date, that is giving consumers something tangible.

Cause for me, if I'm going to take 40Gb at home (and if it was on offer I would 100% want it) I need to have some kind of full specification with regards to usage or its pointless. I'm not going to buy a Ferrari and them say well it'll do 200 MPH but at some point if you drive it too much we might take the car back, we won't tell you if you're driving it too much though, you'll just have to guess when.

Anything you could do with 1Gb-2Gb is absolutely dwarfed when we get into 10Gb-100Gb territory. We're moving from streaming some videos and downloading large games to running a business from your house, one with very high levels of traffic.

I can only assume the price will be very high and so this niche sized plan is going to appeal the most to two kinds of people. Those who just want the fastest and have the money and won't actually use nearly any of it, lets call them the 1%'ers. And those who think it over, plan ahead for what they can do with it, perhaps stop renting some servers online and bring those in-house instead.

I run an online business, I have a great many servers with connections ranging from 1Gb to 100Gb. For redundancy I wouldn't get rid of them if I had this line at home, but I would certainly add a server to my home to use as part of my global infrastructure and do tens of terabytes a month minimum. I'd have to just to make the high cost worth it and I wouldn't do that if I don't know how much of it I can use.

So overall this is just a minor thing like I said, I think it's wonderful I can't wait to see people post their installs and show us some fantastical hardware setups and speedtest results.
 
Top
Cheap BIG ISPs for 100Mbps+
Community Fibre UK ISP Logo
150Mbps
Gift: None
Virgin Media UK ISP Logo
Virgin Media £22.99
132Mbps
Gift: None
Vodafone UK ISP Logo
Vodafone £24.00 - 26.00
150Mbps
Gift: None
NOW UK ISP Logo
NOW £24.00
100Mbps
Gift: None
Plusnet UK ISP Logo
Plusnet £25.99
145Mbps
Gift: £50 Reward Card
Large Availability | View All
Cheapest ISPs for 100Mbps+
Gigaclear UK ISP Logo
Gigaclear £17.00
200Mbps
Gift: None
Community Fibre UK ISP Logo
150Mbps
Gift: None
Virgin Media UK ISP Logo
Virgin Media £22.99
132Mbps
Gift: None
Hey! Broadband UK ISP Logo
150Mbps
Gift: None
Youfibre UK ISP Logo
Youfibre £23.99
150Mbps
Gift: None
Large Availability | View All
Sponsored Links
The Top 15 Category Tags
  1. FTTP (6024)
  2. BT (3638)
  3. Politics (2720)
  4. Business (2439)
  5. Openreach (2405)
  6. Building Digital UK (2330)
  7. Mobile Broadband (2143)
  8. FTTC (2083)
  9. Statistics (1899)
  10. 4G (1813)
  11. Virgin Media (1762)
  12. Ofcom Regulation (1582)
  13. Fibre Optic (1467)
  14. Wireless Internet (1462)
  15. 5G (1404)
Sponsored

Copyright © 1999 to Present - ISPreview.co.uk - All Rights Reserved - Terms  ,  Privacy and Cookie Policy  ,  Links  ,  Website Rules