Superhub : Next DOCSIS 3 modem - these okay?

[Sephiroth]

Originally Posted by Kushan

The only downside being that if your broadband goes down, you can't phone in to get it checked

That's the trick. Built in reliability & redundancy on all current failure modes.

[Kushan]

I don't know how they could possibly achieve that without essentially laying the network down twice.

[craigj2k12]

I don't know of anyone who doesn't own a mobile phone

[broadbandking]

Originally Posted by craigj2k12

I don't know of anyone who doesn't own a mobile phone

What about the people on PAY AS YOU GO that have no credit.

[BenMcr]

Originally Posted by broadbandking

What about the people on PAY AS YOU GO that have no credit.

Pretty sure if you've got a Virgin Mobile even on PAYG with no credit, 150 is still free

[Ignitionnet]

Originally Posted by Kushan

I don't know how they could possibly achieve that without essentially laying the network down twice.

Your current phone line connects to a single point of failure line card via a single point of failure piece of copper, and from there goes on a fibre link whose resilience to JCB action may or may not be especially strong.

Resilience is supplied by using battery back up power supply for the amplifiers.

That is adequate, alongside ensuring that you've Pre-EQ and the UGS service flows that VoIP will use run at a very conservative modulation or aggressive dynamic upstream modulation profile if it's possible. Given each call requires just 64kbps running them at QPSK is fine.

QPSK with Pre-EQ, etc, etc, means upstream SNRs right down to ~12dB are feasible.

[Kushan]

Originally Posted by Ignitionnet

Your current phone line connects to a single point of failure line card via a single point of failure piece of copper, and from there goes on a fibre link whose resilience to JCB action may or may not be especially strong.

Resilience is supplied by using battery back up power supply for the amplifiers.

That is adequate, alongside ensuring that you've Pre-EQ and the UGS service flows that VoIP will use run at a very conservative modulation or aggressive dynamic upstream modulation profile if it's possible. Given each call requires just 64kbps running them at QPSK is fine.

QPSK with Pre-EQ, etc, etc, means upstream SNRs right down to ~12dB are feasible.

That's all well and good, but historically the broadband has been much less reliable than the phone line (Not a slant on the infrastructure, I think it's just the nature of a much more complicated beast). In all the years I've been with Virgin, I think I've had one phone fault but numerous broadband faults.

As for bandwidth, the problem is when congestion hits. It's fine saying you can run it off QPSK but the total bandwidth drops considerably and people are still going to be downloading stuff. QoS is the answer to that but again, historically, Virgin have never been particularly great at that sort of thing and I'm not confident it'll work well in those situations.

[Ignitionnet]

Originally Posted by Kushan

As for bandwidth, the problem is when congestion hits. It's fine saying you can run it off QPSK but the total bandwidth drops considerably and people are still going to be downloading stuff. QoS is the answer to that but again, historically, Virgin have never been particularly great at that sort of thing and I'm not confident it'll work well in those situations.

The standards for telco over cable are very well established and work fine across millions of installed lines throughout the world.

It isn't possible for it to be affected by congestion, the VoIP takes precedence over everything.

It would only be the VoIP that runs over QPSK, everything else remains as it was.

EDIT: http://en.wikipedia.org/wiki/PacketCable

VoIP services based on PacketCable architecture are being widely deployed by operators:
<snip>
Ziggo - "Ziggo Telefonie" (The Netherlands)
UPC Broadband - "UPC Broadband" (Across Europe)
Unitymedia (UPC Germany)
KabelBW - "Clever Kabel"(Germany)
<snip>

All Liberty Global. VM have staff trained in PacketCable and can make use of the experience of the Liberty architecture group at Schipol, Netherlands.

[alanbjames]

Originally Posted by Sephiroth

That's the trick. Built in reliability & redundancy on all current failure modes.

its like when u ring in to report a fault with your broadband and that automated woman says to check server status online.

[qasdfdsaq]

Originally Posted by Kushan

That's all well and good, but historically the broadband has been much less reliable than the phone line (Not a slant on the infrastructure, I think it's just the nature of a much more complicated beast). In all the years I've been with Virgin, I think I've had one phone fault but numerous broadband faults.

I think that's a misconception primarily based of the fact most people use their broadband more than their phone therefore more likely to notice when it's broken, plus most people blame anything that prevents access to a website on their broadband, whether the problem is remotely related to them or not.

A fault with the line at the other end people blame on the other person having no mobile reception, or the other person's telephone provider because your own must be working because you get a dial-tone and you can call everyone else. A broken server or CDN at the far end, a broken DNS server or a routing cock-up at a border gateway though, people will blame that on a broadband fault, even though you get a dial-tone and can still call everyone else.

Inter-provider gateway faults on POTS are just as common IMO, though the lack of dynamic routing protocols somewhat mitigates the number of possible failure modes.

As for bandwidth, the problem is when congestion hits. It's fine saying you can run it off QPSK but the total bandwidth drops considerably and people are still going to be downloading stuff. QoS is the answer to that but again, historically, Virgin have never been particularly great at that sort of thing and I'm not confident it'll work well in those situations.

Bandwidth for cable voice services will be completely separate and isolated from broadband. It would be just as impossible for broadband congestion to affect telephone service as it is now.

---------- Post added at 12:15 ---------- Previous post was at 12:12 ----------

Originally Posted by Ignitionnet

All Liberty Global. VM have staff trained in PacketCable and can make use of the experience of the Liberty architecture group at Schipol, Netherlands.

VM have staff trained in user support and can make use of the experience of many of their better agents, seen here and elsewhere, but seem to still prefer deferring to the Indian subcontinent...

---------- Post added at 12:16 ---------- Previous post was at 12:15 ----------

Originally Posted by broadbandking

What about the people on PAY AS YOU GO that have no credit.

Not much different to the people taking broadband only right now and don't even have a phone line, VM being the only network on which that's possible.

[BenMcr]

Originally Posted by alanbjames

its like when u ring in to report a fault with your broadband and that automated woman says to check server status online.

Which you can do from an internet enabled device such as a Smartphone.

[Kushan]

I'll believe it when I see it! I'm not convinced it'll go smoothly at all but we'll see.

[craigj2k12]

Originally Posted by broadbandking

What about the people on PAY AS YOU GO that have no credit.

I'm sure those people could spare a fiver out of their jobseekers allowance

[Ignitionnet]

Originally Posted by Kushan

I'll believe it when I see it! I'm not convinced it'll go smoothly at all but we'll see.

I have absolutely no doubt there will be initial issues. That's why there'll be trials in labs, then smaller scale in the field, beforehand, to ensure the combination of CPE and CMTS hardware and firmware behaves

It's not some voodoo magic that's unique to VM. Comcast, Cogeco Cable Canada, Rogers Canada, Time Warner, bits of the UPC family, etc, have been selling this stuff since 2005.

I believe that the UPC family is going to be slowly but surely evolving to using the same CPE, CMTS, CCAP, etc, for obvious reasons, and when that happens the most profitable and revenue rich segment of the business, VM, can have smaller sister companies run their scale testing for them

[Kushan]

Originally Posted by Ignitionnet

I have absolutely no doubt there will be initial issues. That's why there'll be trials in labs, then smaller scale in the field, beforehand, to ensure the combination of CPE and CMTS hardware and firmware behaves

It's not some voodoo magic that's unique to VM. Comcast, Cogeco Cable Canada, Rogers Canada, Time Warner, bits of the UPC family, etc, have been selling this stuff since 2005.

I believe that the UPC family is going to be slowly but surely evolving to using the same CPE, CMTS, CCAP, etc, for obvious reasons, and when that happens the most profitable and revenue rich segment of the business, VM, can have smaller sister companies run their scale testing for them

Fingers crossed! I actually really like VoIP, we use it in the office and it's fantastic - but we have a leased line, so it doesn't tend to go down. If Virgin can pull it off, it'll be a bit of a game changer as they can do away with line rental.