Thursday, October 12, 2017

10 Gbps Symmetrical Cable Broadband Speeds are on the Horizon -

&nbspstory category

10 Gbps Symmetrical Cable Broadband Speeds are on the Horizon

The latest upgrade path for the DOCSIS 3.1 cable standard promises a future where cable broadband upstream speeds aren't quite so pathetic. Traditionally, even top shelf cable operators like Comcast struggle to offer anything faster than 35 Mbps on the upstream side of the equation due to restrictions in the DOCSIS standard. But Cable Labs this week formalized an upgrade path for the standard that paves the way toward a future where cable operators are able to offer symmetrical speeds as fast as 10 Gbps.

Click for full size
More specifically, the Cable Labs announcement states the group has completed the Full Duplex Data Over Cable Service Interface Specification update to DOCSIS 3.1 that should make symmetrical speeds possible over cable lines.

Cable's upstream has long been relegated to a limited slice of bandwidth (5 MHz to 42 MHz) referred to as a "low split." To dramatically increase upstream cable speeds, cable operators have been exploring a "mid-split" that would bump the ceiling to 85 MHz, or a "high-split" that would push it to 200 MHz. Full duplex technology would eliminate the need for these splits entirely.

"Current DOCSIS networks have to juggle available upstream and downstream traffic," CableLabs Research and Development VP Belal Hamzeh says of the upgrade. "Full Duplex DOCSIS technology supports multi-gigabit symmetric services by enabling concurrent transmissions in the same spectrum, providing the ability to increase the upstream capacity without sacrificing downstream capacity. This has the potential to greatly improve network efficiency and, in turn, customer experience."

The full duplex version of DOCSIS 3.1 "offers high speeds over the existing infrastructure and is less expensive to deploy than fiber, while still maintaining backwards compatibility with previous generations of DOCSIS technology," the organization said.

Granted you may still be waiting a while before you see these faster upstream speeds, especially if your cable provider appears to be stuck in neutral when it comes to cable upgrades (**cough, Charter**). Commercial deployments of the standard aren't expected for another few years, and how much competition your cable provider sees will dictate how quickly the standard update is deployed.

Most analysts predict the initial wave of full duplex DOCSIS 3.1 upgrades should start appearing in the later half of 2019 or early part of 2020.

karpodiem
Hail to The Victors
Premium Member
join:2008-05-20
Detroit, MI

How about we get upload speeds above 10Mbit first?

Most of the plant is capable of it. So ask yourself - why is upload limited to 10Mbit for most providers?

$$$

And I genuinely feel they are worried about people renting out Slingboxes.

Tchaika
join:2017-03-20
New Orleans, LA

Re: How about we get upload speeds above 10Mbit first?

said by karpodiem:

Most of the plant is capable of it. So ask yourself - why is upload limited to 10Mbit for most providers?

Because it's not. 4 x 27Mbits = 108Mbits. Do the math on the contention for a typical HFC node. Cox sells 30Mbits here, which is basically giving one customer the right to try and use ~30% of the available bandwidth, and it can only work if nobody actually tries to use what they pay for with any measure of consistency.

Before you say it, eight channel modems are mostly marketing fluff. With existing OOB uses, a 42MHz ceiling, and 6.4MHz wide channels, four is the practical limit. You could get eight, with 3.2MHz wide channels, but that would be self-defeating.

A better question is why the industry is wasting all this money on 3.1 and future DOCSIS standards when they're already struggling with noise on their coax plants and already have fiber pretty damned close to most of their customers. Seems to me the money could be better spent on FTTP than DOCSIS 3.1/Node+1, but what do I know......

Re: How about we get upload speeds above 10Mbit first?

said by Tchaika:
said by karpodiem:

Most of the plant is capable of it. So ask yourself - why is upload limited to 10Mbit for most providers?

Because it's not. 4 x 27Mbits = 108Mbits. Do the math on the contention for a typical HFC node. Cox sells 30Mbits here, which is basically giving one customer the right to try and use ~30% of the available bandwidth, and it can only work if nobody actually tries to use what they pay for with any measure of consistency.
But any one subscriber on a node probably would only use the full speed of 30 Mbps for a brief period (unless performing an occasional online backup, etc.), and would not be "using ~30% of the available bandwidth" all the time.

Just like not all subscribers on a node fully use the 100 to 200 Mbps download speed of their links all the time. The total bandwidth of the typical 24 channel D3.0 downstream arrangement wouldn't handle that either!

I've seen some cable MSOs advertise 50 Mbps upload speeds on their highest tiers right now.

Tchaika
join:2017-03-20
New Orleans, LA

Re: How about we get upload speeds above 10Mbit first?

said by telcodad:

But any one subscriber on a node probably would only use the full speed of 30 Mbps for a brief period (unless performing an occasional online backup, etc.), and would not be "using ~30% of the available bandwidth" all the time.

I understand the theory but the reality is that their ego is writing checks their body can't cash.

I never signed up for that tier because I've never seen it actually deliver the promised speed, not even for bursts. I've watched Cox techs run speedtests on their meters and not manage to reach ~30Mbits, they usually top out at 24-25Mbits (during business hours, so it's probably worse during peak times), which is in line with what I've seen at customer sites that have that tier.

10Mbits upload is still relying on over-subscription (~10% of the available bandwidth) but at that level of over-subscription you're actually going to be able to deliver what your customers are paying for more often than not.

For folks who care about QoS, having your available upstream fluctuate all over the place is a recipe for problems. I want to control when/how my packets queue, so my jitter/latency sensitive applications function properly, and I can't do that if my provider can't consistency deliver the upstream I'm paying for.....

to Tchaika
Overall deployment and maintenance cost is the biggest driver.

Its a 100% no brainer that a coax drop is cheaper, faster, easier, and more desirable to deploy to a home from the plant than a fiber drop.

Right now my at bulk price (hundreds of thousands of feet at a time) for a flat fiber drop vs a coax cable drop is double. I'm $0.05 per foot on cox cable vs $0.11 per foot on fiber.

To get a proper splice, it also require a several thousand dollar fusion splicer (those mechanical end splices are junk for reflection, and not cheap either at $11 each, mechanical or fuse). My coax ends are $0.17 each.

Fiber taps? Several hundred each. Coax taps? $30

Time taken in order to do do each drop is fairly higher either. We do a mix of them both as we do fiber to businesses right now. Spanning a single pole drop to a new house is about a 30 minute process, end to end all terminated. Its about 2 hours for a fiber, if the sun, moon, and stars all align.

Same rules apply for both, you won't be splicing the old drop back together if it gets tore down. We don't do it with coax anymore because the mid splices are usually where we have issues. You won't do it with fiber because its more expensive to aerially fusion splice the break (remember, no slack to do it on the ground) than it is to replace it all. However! If i need to get you on in a pinch I'll mid splice the coax. Your fiber? Its staying out till a new drop can be placed.

So ignoring fixing the breaks, lets look at a drop cable breakdown: At an average of 300ft per drop in this area, over 1000 homes and a $25 an hour tech:

$105 per fiber drop, or $105,000 per thousand subs.
$27.59 per coax drop, or $27,590 per thousand subs.

That $77,410 per thousand I bet will go a long way towards a FDX deployment, especially if we are talking using existing node fiber because of CWDM wavelength deployment on nodes. And end of day it would be creating a product that would be insanely easier to maintain to the home.


Zenit
Premium Member
join:2012-05-07
Purcellville, VA
10.2 0.8
·Verizon DSL
·Comcast XFINITY
·T-Mobile
to karpodiem
It's a bit of a stretch to say a typical 3 upstream channel plant can handle massive upload, at least with D3.0 64QAM modulation upload channels. Technical limitations are real.

30.72Mbps bandwith of a single US channel at 64 QAM x 3 = 92.16Mbps total capacity shared across all users on the node or CMTS upload line card if nodes are combined.

The HFC nodes in my area (860mhz Comcast owned plant) have roughly 55 people on average per node. In order to not oversubscribe the upload would need to be capped at 1.675Mbps for each customer. Even with the low upload speeds they provide today the system is still oversubscribed.

It's honestly amazing that HFC cable provides a low latency path given the over subscription ratio. D3.0 upload clearly wasn't meant for everyone on a node to be uploading mass quantities of information at all times.

The solution is DOCSIS3.1. Even a single OFDM upload channel crammed into the small space would greatly increase available upload bandwidth. It will be some time before they sacrifice one or two D3.0 upload channels for a D3.1 upload channel. There are too many old D3.0 and D2.0 modems out there that can't decode OFDM.

What is delaying D3.1 upload? Technical challenges (needs a clean return path), and money. Lot's of money involved in swapping out a majority of the modems.

Tchaika
join:2017-03-20
New Orleans, LA

Re: How about we get upload speeds above 10Mbit first?

said by Zenit:

The solution is DOCSIS3.1.

You spelled 'FTTP' wrong......

Cox has fiber already passing nearly all of their customers in this city. The only ones that might not be passed on those are short dead-end streets.

Their plan is to spend a massive amount of money on 'Node+1', basically splitting the network down to the point where you've got the HFC node, one amplifier, and that's it. The current standard is Node+3 or even +4, depending on the deployment. This is why they have fiber everywhere, they need it for future node splits.....

Node+1 will help with contention ratios and it will help with their noise ingress problems, but there comes a point where you have to ask yourself if chasing these diminishing returns is really worth it, or if they should just bite the bullet and go to FTTP. FTTP would cost more today, but it would cut down on maintenance expense tomorrow, and future-proof them for the day after tomorrow.

Re: How about we get upload speeds above 10Mbit first?

I think Altice figured that out already and Verizon some time ago.
Keep in mind that they have to fully depreciate the assets before this becomes realistic.


Zenit
Premium Member
join:2012-05-07
Purcellville, VA
10.2 0.8
·Verizon DSL
·Comcast XFINITY
·T-Mobile
to Tchaika
Cable providers are finding out that DOCSIS3.1 is way more resilient on the downstream than originally expected. Due to the diminishing revenue of the Cable TV unit I can't see most MSO's plunking down massive cash on FTTP to every customer any time soon. We will see a phased approach. Node+1 will be the standard.

I think MSO's are mostly worried about the expense of laying a new fiber drop and installing an ONT at each home. Most of the hard work has been figured out by Verizon already, so many of the installation pitfalls are no longer valid.

Comcast seems to be fairly ambitious with regards to new plant. They are letting people who really want FTTP buy it from them at a cost well below what an actual Metro Ethernet circuit would cost. At least there is the option.

Tech like "Deep Fiber Solutions" which lets MSO's eject the core of plant coax and turn it into conduit has the potential to make FTTP deployment easy, at least easier than what VZ had to do.

»www.youtube.com/watch?v= ··· USGLCi0M
Tchaika
join:2017-03-20
New Orleans, LA

Re: How about we get upload speeds above 10Mbit first?

said by Zenit:

I think MSO's are mostly worried about the expense of laying a new fiber drop and installing an ONT at each home. Most of the hard work has been figured out by Verizon already, so many of the installation pitfalls are no longer valid.

That's a valid concern, but I can't help but think about my shitty experiences with Cox, all owed to noise issues on their plant, none of which would be an issue with FTTP. It absolutely blows my mind how bad the noise issue is here, particularly in my neighborhood, but really in the whole damn market to varying degrees.

They could fix their coax plant, but that would also cost money, because they'd have to get rid of their shortcut taking contractors and rely on their more competent in-house techs that are paid by the hour rather than by the job. They'd also have to hire more people to do preventive maintenance on the plant, which they regard as a waste of money, because they're not out doing installs.

Re: How about we get upload speeds above 10Mbit first?

Noise is a lot less of a concern when you have a 1/5th of the plant that you used to have feeding the node your on. A reduction in plant size per node gains your overall noise resiliency due to less area for ingress to enter and compound into an overall noise floor reduction.

And on top of that, its easier to track and correct plant issues when your going deeper with the node count. All of a sudden what used to be 10's of miles of plant per node is now a mile. A few drops and adds of main line gear and you have it narrowed down and tagged out.

With less customers per node noise is also affecting less people at once as well.


djrobx
Premium Member
join:2000-05-31
Valencia, CA
to karpodiem
The plant may be capable of 20-35mbps but that doesn't necessarily mean it can support all of its customers with such speeds.

This improved full duplex technology means they could greatly increase upload speeds with less risk of congestion.

Optimum Online offers higher upload speeds with existing DOCSIS, but I believe their plant was engineered with fewer homes per node than a typical cable provider.


Zenit
Premium Member
join:2012-05-07
Purcellville, VA
10.2 0.8
·Verizon DSL
·Comcast XFINITY
·T-Mobile

Zenit

Premium Member

Re: How about we get upload speeds above 10Mbit first?

And even with a lower number of homes per node, you can still run into congestion on the upload fairly easily if you have a neighborhood full of tech-heavy households.

Provisioning speeds on DOCSIS is a delicate balancing act. When Comcast bumped 50Mbps to 105Mbps in my area it caused massive congestion because our old 8 channel bonding Arris C4 CMTS couldn't cope with the load anymore, lack of ports meant nodes were combined. Doubling the speed a majority of the customers can download at created an awful user experience, think sub-3Mbps speeds at peak hours.

They replaced the Arris C4 fairly quickly with an E6000 but the wait time was painful.



How about ..

Let's block ads! (Why?)


see source

No comments:

Post a Comment