HFC - When will we get better upload speed?

Forums › One New Zealand (including Vodafone, WxC, Farmside) › HFC - When will we get better upload speed?

DonGould

3892 posts

Uber Geek

#115702 3-Apr-2013 22:15

10mbits of upload speed is just pathetic given what this technology can do.

UFB is doing 100/50

I don't see why HFC can't have 100/100 given that it's a DOCSIS3 network.

Why is the speed so poor?

Promote New Zealand - Get yourself a .kiwi.nz domain name!!!

Check out mine - i.am.a.can.do.kiwi.nz - don@i.am.a.can.do.kiwi.nz

1 | 2

19282 posts

Uber Geek
Inactive user

#792750 3-Apr-2013 22:19

UFB is not the same as the cable HFC network and we can't speak about future plans or changes on a public forum and you are asking wayyyyyyy to early,

MyHeritage

Shop MyHeritage and uncover your origins and find new relatives with a simple DNA test. (affiliate link).

freitasm

BDFL - Memuneh
76390 posts

Uber Geek

Administrator

ID Verified

Trusted

Geekzone

Lifetime subscriber

#792768 3-Apr-2013 22:31

Don asked why the speed on DOCSIS 3 services are limited to 10 Mbps uploads when the technology itself allows ten times more.

This is obviously for historical reasons, but it's a good question. It's also a good feedback. We are seeing a shift from consumption to creation - photo uploads to Facebook, Pinterest, Flickr and Picasa, online backups, P2P and other services that use upload streams.

It's natural then to ask the hard question: does it make sense to limit upload speed in a fast plan like that?

Please support Geekzone by subscribing, or using one of our referral links: Dosh referral: 00001283 | Sharesies | Goodsync | Mighty Ape | Backblaze

freitasm on Keybase | My technology disclosure

johnr

19282 posts

Uber Geek
Inactive user

#792773 3-Apr-2013 22:36

That is why I said " future plans or changes " what Don is asking about is change to the network for Upload speed

I have Zero idea why it's limited to 10Mbps Up-link Steve Biddle would have better chance of knowing than me

freitasm

BDFL - Memuneh
76390 posts

Uber Geek

Administrator

ID Verified

Trusted

Geekzone

Lifetime subscriber

#792776 3-Apr-2013 22:39

Well, answer my question then... Does it make sense limiting speeds in a plan like that?

Please support Geekzone by subscribing, or using one of our referral links: Dosh referral: 00001283 | Sharesies | Goodsync | Mighty Ape | Backblaze

freitasm on Keybase | My technology disclosure

johnr

19282 posts

Uber Geek
Inactive user

#792778 3-Apr-2013 22:41

freitasm: Well, answer my question then... Does it make sense limiting speeds in a plan like that?

Would depend on the reason why it's capped at that up-link speed and I edited my post above

DonGould

3892 posts

Uber Geek

#792788 3-Apr-2013 22:55

I'm not interested in the historic rhetoric. I can speculate on that and likely hit enough of the nail on it's head.

MF is right, we have moved into an age of content creation and cloud computing where we want to push content to SAAS quickly.

http://en.wikipedia.org/wiki/DOCSIS#Speed_tables

As the upload channels will only do ~128mbit, 60mbit is as much as we can realistically expect as you need to allow at least double over head for a shared service to not get congested.

But yes, 60mbit is a far cry from 10mbit.

I hear and accept John's comment that I'm likely asking a bit early in the peace, but at the same time they've had 6 months now to start to get this stuff sorted out, and likely a lot longer in due diligence.

While I know it's not good to kick the gift horse in the mouth, having just got 50% more value with a 15% saving, it still has to be remembered that we're still only seeing 10% of the value in these plans that our neighbours in Australia are seeing on their networks.

Promote New Zealand - Get yourself a .kiwi.nz domain name!!!

Check out mine - i.am.a.can.do.kiwi.nz - don@i.am.a.can.do.kiwi.nz

johnr

19282 posts

Uber Geek
Inactive user

#792792 3-Apr-2013 23:02

Don you can't expect the world to change in 6 months, This was a big acquisition and I am sure the top of the list would not have been to change the up-load speed on the cable network,

6 months is very little time in the grand scale of things

DonGould

3892 posts

Uber Geek

#792799 3-Apr-2013 23:24

johnr: Don you can't expect the world to change in 6 months, This was a big acquisition and I am sure the top of the list would not have been to change the up-load speed on the cable network,

6 months is very little time in the grand scale of things

Actually John, in the world of the internet, I can expect change that fast.

Don't get me wrong here, or read to much into what I'm saying.

I simply asked when we can expect more speed and expressed that 10mbit is poor given what the technology can do.

Clearly from your comments you also agree that it's poor, but also a little to soon for you to be able to comment on when we might see change, and I accept that.

Where would you put upload speed on the list of things to do? (...and I appreciate that might not be something you can comment on as you might actually have every idea where that sits but is not appropriate to say.)

Personally I'd have thought that would be very high on the list for VF as it is something I think we all really really want and really could paint VF in a very good light in the market as it really would deliver a real point of difference that the market doesn't currently have.

I also wonder what people might be willing to pay for more upload speed.

This week I saved $20 a month, but you can have that back again in return for 20 more mbits (ie 100/30).

I think I'd be willing to pay $1 per mbit. So if I could have 100/50 I'd be willing to pay $40 dollars more a month for that value.

But don't get me wrong. I do understand the impact that might have on the BRAS and network configuration in general to suddenly see 5 times the volume of data flowing though the BRAS that way.

I get that essentially means that 50/50mbit connections are instantly possible and that could have quite an impact on local load. It might also have a significant impact on the NNI links that other providers then need to purchase off VF to accept the volumes of traffic that might start being thrown at them.

/random thoughts :)

Promote New Zealand - Get yourself a .kiwi.nz domain name!!!

Check out mine - i.am.a.can.do.kiwi.nz - don@i.am.a.can.do.kiwi.nz

DoomlordVekk

129 posts

Master Geek

Trusted

#793325 4-Apr-2013 16:40

Don, the problem you may be experiencing is "Unrealistic expectation". The HFC/Cable networking technology is by it's nature and the most common deployment scenario, asymmetric and very residentially focused.

The downstream traffic is pumped by box technology that cost hundreds of thousands of $$.

The upstream traffic is pumped by your cable modem that costs less than US$150 each.

The modulation depths downstream and symbol density is often 8 times more in the downstream compared to the upstream.

HFC/DOCSIS is heavily used as residential systems because it cost effectively delivers the higher downstream speeds that 85% of the market looks for, not the symmetrical traffic options that something like Metro Ethernet or the Private IP services that are business class services.

So, you may be trying to apply symmetrical thinking to what is inherently asymmetric.

All I can suggest is if your data needs upstream have grown to a point where you've outgrown the Cable network, you may need step up to a more correctly dimensioned business service.

"Customers don’t expect you to be perfect. They do expect you to fix things when they go wrong." Donald Porter – British Airways

The views expressed here are my own and are not reflective of other organisms or organisations.

DonGould

3892 posts

Uber Geek

#793335 4-Apr-2013 16:53

DoomlordVekk: Don, the problem you may be experiencing is "Unrealistic expectation". The HFC/Cable networking technology is by it's nature and the most common deployment scenario, asymmetric and very residentially focused.

The downstream traffic is pumped by box technology that cost hundreds of thousands of $$.

The upstream traffic is pumped by your cable modem that costs less than US$150 each.

The modulation depths downstream and symbol density is often 8 times more in the downstream compared to the upstream.

HFC/DOCSIS is heavily used as residential systems because it cost effectively delivers the higher downstream speeds that 85% of the market looks for, not the symmetrical traffic options that something like Metro Ethernet or the Private IP services that are business class services.

So, you may be trying to apply symmetrical thinking to what is inherently asymmetric.

Interesting, thanks DV.

I really don't get why wikipedia publishes the figures that I linked to then.

You're saying that the 128mbit up stream line rate is just unrealistic?

What is a realistic line rate on the technology we have in the network?

DoomlordVekk: All I can suggest is if your data needs upstream have grown to a point where you've outgrown the Cable network, you may need step up to a more correctly dimensioned business service.

As far as I'm aware, no one actually has infrastructure anywhere near me that will let me do that at present... roll on UFB I guess.

I'm starting to see why others were keen to see UFB over build HFC now.

I wonder if these same constraints also impact GPON and if we'll see upload limits stop at 50mbit while we see the download speeds grow to match the link capacity?

Promote New Zealand - Get yourself a .kiwi.nz domain name!!!

Check out mine - i.am.a.can.do.kiwi.nz - don@i.am.a.can.do.kiwi.nz

sbiddle

30853 posts

Uber Geek

Retired Mod

Trusted

Biddle Corp

Lifetime subscriber

#793341 4-Apr-2013 17:01

DonGould:
I wonder if these same constraints also impact GPON and if we'll see upload limits stop at 50mbit while we see the download speeds grow to match the link capacity?

You can't compare GPON and DOCSIS3. GPON 100/50 dimensioning exists so a 24 way split isn't ever contended, so unless this is relaxed to somebody who wants to offer faster speeds and cause contention on the network upload speeds will never exceed 50Mbps.

techo

46 posts

Geek

#793446 4-Apr-2013 19:58

DonGould:

Interesting, thanks DV.

I really don't get why wikipedia publishes the figures that I linked to then.

You're saying that the 128mbit up stream line rate is just unrealistic?

DOCSIS 3.0 upstream speeds of 100Mbps can be achieved in a lab under ideal conditions. What can be achieved in practice comes down to HFC plant design, infrastructure, and plant condition.

Disclosure: I work for Vodafone.

sbiddle

30853 posts

Uber Geek

Retired Mod

Trusted

Biddle Corp

Lifetime subscriber

#793456 4-Apr-2013 20:06

techo:
DonGould:

Interesting, thanks DV.

I really don't get why wikipedia publishes the figures that I linked to then.

You're saying that the 128mbit up stream line rate is just unrealistic?

DOCSIS 3.0 upstream speeds of 100Mbps can be achieved in a lab under ideal conditions. What can be achieved in practice comes down to HFC plant design, infrastructure, and plant condition.

Disclosure: I work for Vodafone.

DOCSIS3 is technically capable of "infinite" speeds in both directions as speed is only limited by capabilities of channel bonding. Right now there is no mass market DOCSIS3 kit that can utilise more than 4 bonded upstream channels (which is ~100Mbps)

As to why TCL didn't operate bonded upstream channels - that's only something they can answer publically.

DonGould

3892 posts

Uber Geek

#793465 4-Apr-2013 20:21

sbiddle: As to why TCL didn't operate bonded upstream channels - that's only something they can answer publically.

Why TCL didn't do it doesn't really interest me.... when VF will do it, does interest me. :)

I am interested to know more about what the kit in our network can do and what it will take to get the most legs out of the puppy.

I'm interested to know when we can start to expect to see the sort of value that Australians get.

I'm also interested to hear what others would like and what they'd be prepared to give for it.

I've outlined my opening offer, what gold would others consider handing over for more value?

Promote New Zealand - Get yourself a .kiwi.nz domain name!!!

Check out mine - i.am.a.can.do.kiwi.nz - don@i.am.a.can.do.kiwi.nz

DoomlordVekk

129 posts

Master Geek

Trusted

#793507 4-Apr-2013 21:28

sbiddle:

DOCSIS3 is technically capable of "infinite" speeds in both directions as speed is only limited by capabilities of channel bonding. Right now there is no mass market DOCSIS3 kit that can utilise more than 4 bonded upstream channels (which is ~100Mbps)

As to why TCL didn't operate bonded upstream channels - that's only something they can answer publically.

Regretably, thats all not quite true. The DOCSIS3 standard is build on a frequency range of approx 600MHz of downstream spectrum and (if you're lucky), 80MHz of upstream spectrum. That effectively limits the (A) number of channels and (B) bandwidth of channels.

Upstream is the rarest of comodities in the cable world. To increase available upstream channel capacities to customers, you basically have to take your existing fibre nodes(suburbs serviced by a ONT and it's coaxial runs), cut them in half and build 2 new nodes, with all the fibre, coaxial and equipment costs that go with that.

To get Don's 108Mb in the Upstream would mean dedicating 4 entire channels to each customer, maybe with some stat muxing gains, that would let you share those channels amongst 20-30 customers.
Cards for CMTSs that provide ~20 upstream channels are listed on eBay for between US$22,000 and US$91,000! You do the maths...

Likewise the silicon processing engines in the cable modems have bonding/multiplexing limitations that have to be taken into account (how vastly clever a piece of radio technology are you normally able to buy for under US$150?).

If you have a sleepless night or 2 thousand, you could have a read of the CableLabs standards.

DOCSIS3.1 is in the process of being rationalised right now, it looks really interesting and will promise that next step in throughput and technology but it's still 15-18 monthe before we see the first leading and bleeding equipment being released by the likes of Motorola, Arris and Cisco.

Fibre is always going to win over copper, very few countries would be looking to deploy HFC networks vs *PON, given the more greater bandwidths available and far lower cost of maintenance just to pick 2 points out.

Channels bonding is alive and well and delivering excellent results in the Vodafone HFC networks, anyone on cable, with a Warpspeed plan can confirm it themselves with a bit of googling.

Enjoy! :)

1 | 2