Virgin Media Community

[Andrew-G]

@ToffeeSurprise DOCSIS is rubbish for gaming. 

That's not entirely true.  FTTP is measurably better than HFC for raw latency, but comparing good quality HFC and FTTP connections you'd have to be truly exceptional to tell the difference in real world use.  The difference in minimum latency is about 3-5ms, as is the average and (again, on a good quality DOCSIS connection) there are going to be a few DOCSIS peaks to 28ms.  Since human reaction time to a screen image is about 250ms or worse, the difference between a (possibly) solid 18ms FTTP connection and a 28ms DOCSIS peak of 10ms is immaterial.  But even if the latency differential is higher, it doesn't necessarily matter very much until it's approaching 40ms.

Whilst some gamers fixate on ping, FPS and refresh rates, the truly skilled players still do very well on slow connections and slow machines, because the real differentiator is intuitive anticipation.  Think about international gaming where some players are having to suffer long distance latency penalties, yet still trample the peasants.  I see this often on CSGO, as I'm repeatedly slaughtered by skilled Ruskies who have a 40ms distance related latency added to whatever their local connection has, and the switching and connection latency on the gaming server.  If my latency came down to FTTP standards it would increase the differential in my favour, and still make not a blind bit of difference to the outcome.

Fast reaction times are important, good coordination helps, but aren't enough, and quality equipment is helpful but can't make a good player a great player.  That intuitive anticipation in different contexts is what makes the difference between the best pilots, racing drivers, goalkeepers, tennis players etc, because they react to something before it happens.  This is very obvious with top flight goalkeepers defending a penalty, where you can see (and the physics dictates) that they have to both have committed to where the ball is going and have started to move before the ball is kicked.

We'd all like better and more consistent latency from HFC, and everybody recognises that it is a legacy technology that will need sooner or later to be replaced, but unless there's a problem then DOCSIS/HFC is more than good enough for gaming.  Obviously if there's over-utilisation, or noise or power issues then HFC can be very poor indeed, and VM don't always help themselves on those matters.  But bear in mind that FTTP is for the most part in its early roll out, the FTTP equipment is generally under-utilised and without any traffic stress.  Only time will tell if that remains the case.

[Harvey_H]

Reusing CPE that uses coax isn't really our choice after all though. It is a decision by Liberty Global. Just like how the Hub 4 and Hub 3 are used across Europe.

[Roger_Gooner]

---

@ToffeeSurprise wrote:

Using RFoG and converting to coax to reuse the CPEs is a good way of reducing short term CapEx since you don't need new equipment, but it creates a much longer term problem by tying Virgin Media into equipment, or requiring costly rework in customer homes. Because the ONT has to convert from RFoG to electrical, and the CPE has to use coax connections this imposes limits because there isn't much hardware that does this. Compare this to a conventional FTTP GPON ONT where it's presented to the customer over Ethernet, and basically any router capable of PPPoE can be used. Swapping ONTs for upgrades is trivial.

---

RFoG is an entirely sensible way of introducing FTTP whilst leveraging the major investment in the current network. In fact I would go further and say that it would be ridiculous for VM to have gone to a PON solution immediately on its FTTP network. In time we could have RFoG for TV and EPON for broadband, requiring a change to the ONT and hub. Later, RFoG will proably be gone as well. In IT it's very common for improvements to be made by a series of interim products and services whilst making use and money from previous investments.

[ToffeeSurprise]

Andrew, by all accounts my HFC connection with Virgin Media is good, certainly about as good as you'll get in the UK. Now either you're saying that even Virgin Media's 'good' HFC plant is dreadful, or you believe this is actually good for gaming.

I can tell the difference between the two blind. You can swap me between the two, and I'll be able to tell you without even looking at ping. Even a simple web page request (Wikipedia) has gone from around 2 seconds to load to about 0.1, literally the time it takes for me to blink - despite my VM connection having a faster line rate downstream. Yes, I can and will prove this with a video if needed. The joys of eliminating DOCSIS bufferbloat and having a better bandwidth-delay product.

Do I notice that my ping is constant to the server, at any time of day? Yes.

Do I notice that my latency to EU has gone from mid 30s to 10ms (~66% reduction)? Yes, because I now get kills against these players more in games with a low time to kill. Latency to London from 20ms to 5ms (75% reduction).

Do I notice that I get much less BS because my "I just shot you in the head" packet didn't randomly get delayed for 200ms resulting in far more consistency? Yeup.

Do I notice that I no longer get 100-200ms latency spikes playing WoW? Yes.

Do I notice that I'm no longer getting 60ms to some EU servers due to LG's sup par routing, and I'm not being kicked from servers after a few hours because latency has shot up above the threshold? Yes.

Do I notice that latency to the US has dropped from about 100ms to 80ms on the East coast, and to mid US has dropped by 30ms to 100ms making an entire server region viable for gaming? Yes.

I played a game of Halo a few months ago against a very skilled player. Our skill levels were mostly even, but he has a 120ms advantage. Which meant that basically every single time he got the kill, I might as well have been shooting at air.

Latency does count for gaming more than you think. The complete jack of jitter certainly does.

And to reiterate, this is on one of the best VM HFC connections in the country, with a BQM I suspect many here would murder for.

Night and day.

Can VM get their network to a state where the differences are academic? Yes. Is that the case now? Not a chance.

[Adduxi]

Just jumping in here, I am fortunate enough to have both VM and BT FTTC (for work redundancy) and route all my gaming via the BT circuit. I would also say my VM circuit is good, but despite BT being FTTC it is the better of the two as far as consistent ping etc.  It's also native IPv6 to boot  😉

[Gavdaddy]

Does this mean that in fact VMs FTTP service is in fact FTTC in most cases with copper to the premises?

[Gavdaddy]

Is BT’s FTTP the only true FTTP widely available. Currently have VM Gig1 and it is unusable! At least the last 200m are coax!

[Roger_Gooner]

VM's FTTP is routed via virtual hubs and splitters, Openreach's FTTP is routed via aggregation nodes (with some via DSLAMs in VDSL2 cabinets) and GPON splitters. Both are fibre all the way to an ONT on the exterior wall of the premises where a conversion is done, i.e. to coaxial or Ethernet cable respectively.

[Roger_Gooner]

---

@Gavdaddy wrote:

Is BT’s FTTP the only true FTTP widely available. Currently have VM Gig1 and it is unusable! At least the last 200m are coax!

---

VM delivers Gig1 over both its legacy HFC network and its FTTP network. You are on the HFC network if you have a coaxial cable coming into your home.

[Gavdaddy]

Interesting that none of the cases I have looked at reflect this.

VM ‘FTTP’ has fibre only to a cabinet 250m from the premises which serves several streets - probably more correctly called FTTC.

Openreach provided fibre connection to the ONT inside the premises.

Gavin