Why are the uploads so low on VM?

[danlufan]

Hi All,

i was wondering the reason VM keep their upload speeds so low, even the 50mb is set at 1.5mb upload, this isnt DSL so why set the upload so low?

O2/BE with annex M can reach 2.5mb over an ADSL2+ line and most commonly hit 1.6mb.

Are there ways to have the cap removed, anyone every heard of default profiles being made?

Thanks in advance,

[Ignitionnet]

See the blog in my signature for some explanations of why, and no you can't have the cap removed, that's just how the tiers are at the moment. There is no way to modify them as they are policed at the Virgin side not at your modem.

[danlufan]

See the blog in my signature for some explanations of why, and no you can't have the cap removed, that's just how the tiers are at the moment. There is no way to modify them as they are policed at the Virgin side not at your modem.

Nice blog read, i see you went for the O2 annex-m solution, hopefully Virgin will finish their "trials" soon and give us better upload options.

When i said remove the cap i meant from the virgin system end, i remember the old days when you could create default profiles.

[Ignitionnet]

Nice blog read, i see you went for the O2 annex-m solution, hopefully Virgin will finish their "trials" soon and give us better upload options.

When i said remove the cap i meant from the virgin system end, i remember the old days when you could create default profiles.

In the old days the upstream bandwidth was controlled by the modem and a profile switch on the CMTS would only affect downstream not upstream.

You'd need access to the TFTPs, DHCPs and CMTS to 'do' your upstream - downstream was one command on the CMTS. ;)

[Chrysalis]

ignition do docsis3 cards only come with downstream ports then? so would need 2 of them to bond both up and downstream?

I am sitting here wondering how VM ended up with a docsis3/1.1 combo.

[Ignitionnet]

All hardware that can do DOCSIS 3 can do 2, 1.1 and 1 also.

The cards question isn't so simple.

[|Kippa|]

Do you think they will leave the 5mbit upgrade (if there is one) until they roll out the 100mbit connections at the end of the year. They'll need 5mbit upload rate just for the ack packets.

[Chrysalis]

the question is simple tho.

the question is why arent VM using docsis3 for both upstream and downstream with 50mbit modems, instead of ending up using 1.1 for the upstream.

Docsis3 cards one way only? ie. would need 2 for both ways, or a deliberate decision by VM to strangle capacity.

the answer may not be simple but the question is.

Another question I have is will VM increase upload in areas, where they are already under capacity constraints without increasing capacity first. I think the answer is yes, but some people have more faith in VM then I do, but at the end of the day what gives them more sales will win over. Given the state of the legacy ports in my area, I know first hand VM have no issue with selling services when they cannot deliver.

The ack packets is a very valid point, when I download at 20mbit I upload at around 55-60kB/sec just in ack packets and this is with delayed acks enabled (would be double without nagle algorithm), this is around 2/3rds of the upload capacity given to me by VM.

So what we waiting for is a network wide rollout of docsis2 for 10/20mbit customers and docsis3 for 50mbit customers on the upstream. In addition to that in areas where they only using QPSK they need to allocate double channels to make up for the fact they only half the capacity. I get the impression VM have a lot of work to do before they can boost the upload.

[Ignitionnet]

http://onewayinternet.blogspot.com/2008/09/this-is-technical-science-bit.html

[Chrysalis]

reading :)

---------- Post added at 09:51 ---------- Previous post was at 09:33 ----------

I guess is a bit outdated on the downstream, as now qam256 is higher than 38mbit?

is it easy to tell what network build the area has? I am guessing if its dominated by qpsk upstreams its more likely its an older build?

[Wad_2002]

When have upload rates ever been high?

[Ignitionnet]

reading :)

---------- Post added at 09:51 ---------- Previous post was at 09:33 ----------

I guess is a bit outdated on the downstream, as now qam256 is higher than 38mbit?

Nope I was discussing the legacy network and mentioned assumption. 256QAM DOCSIS is 38M, 256QAM EuroDOCSIS is about 50.

is it easy to tell what network build the area has? I am guessing if its dominated by qpsk upstreams its more likely its an older build?

Yes it's easy. Look at frequencies not modulations. Your area is a brand new overbuild so it's not going to be an older build by default ;)

[danlufan]

Well the fact of the matter is;

Previously the general consumers haven't demanded higher upload rates, so they haven't worked on providing decent speeds.

Now that we use the upload a lot more and people now understand computing terms more the demand has increased, therefore within the next year+ we will see companies focusing a lot more on providing higher uploads. Which in turn reduces the spend on SDSL services.

Money speaks in all of these matters, i just wanted to understand why Virgin hadn't launched their higher upload product yet. Especially now nextgen acces i.e. infinity is starting to come out.

[pip08456]

Well the fact of the matter is;


Money speaks in all of these matters,

Explains why it has taken them so long.

[Ignitionnet]

As I noted previously BT have just announced that as of 30th September they will be offering 15Mbps upstream on their Fibre To The Cabinet service.

[Kymmy]

:cry:

[Chrysalis]

:cry:

haha dont cry.

BT can only give me a line capable of 4-5mbit down.

VM 20mbit :D

VM wins.

[Ignitionnet]

haha dont cry.

BT can only give me a line capable of 4-5mbit down.

VM 20mbit :D

VM wins.

Speak for yourself, I'll in the not too distant future have one capable of 60Mbit or more (when they open the taps) and 15Mbit upstream straight out of the box. This will be a potential option to 2/3rds of the UK.

Makes VM's 1.75Mbps on the 50Mbps service look rather lame to say the least. Game on :)

[Chrysalis]

So BT serve your area well but mine poorly. :)

VM got the move first (enabled in dec 2009), and had a 5 week waiting time on installations for over 3 months as a result.

I expect BT will enable FTTC in my area in a couple of years after VM have all the customers.

on the 2/3rds, what % of uk has already been announced?

BT have missed quite a few cities.

[Ignitionnet]

Probably.

VM may have fixed the congestion issues by then :)

[Chrysalis]

VM isnt perfect but also lets be realistic, you can respect BT's service in my area is a shambles?

How do people like me feel when they providing FTTC to people who are already able to get 15meg on adsl, and leave my area with 50db attenuation lines as a 5mbit'ish service. Then they claim its not viable to rollout FTTC whilst rolling out to lower populated areas.

Also I can report a fault to VM and get a tech out the next day to fix a high power issue, with BT its a fight to get a openreach engineer out whilst avoiding a £125 fee and getting it fixed in a 2 hour slot.

(yes VM staff whilst I have attacked VM, I do feel BT are a lot worse).

VM acknowledged my speeds before I got moved were unacceptable and I had free bb for a bit and did eventually get moved to overlay to get it fixed. BT instead claim long line issues are out of their control and its a case of tough luck.

yes I am *jealous* tho if you have 60/15 :) but in my case I went from 4-5mbit down on a line with sporadic noise issues to 20mbit down so VM will still win in areas like mine.

[Kymmy]

When you live in an area which doesn't even yet have ADSL2 VM is really the only option, just wish the business side would get thier act together :(

[Ignitionnet]

Also I can report a fault to VM and get a tech out the next day to fix a high power issue, with BT its a fight to get a openreach engineer out whilst avoiding a £125 fee and getting it fixed in a 2 hour slot.

Again - swings and roundabouts, I've had to wait a week for VM to fix a telephony issue. Phone was totally dead for a week.

All arbitrary and variable depending on where in the country you are and the workload at the time. If VM's service techs are all busy in your area you'll have a long wait too, and the fees are due to the way Openreach work, if VM are forced to wholesale I've no doubt they'll be doing similar to avoid their wholesale customers loading up their techs with trivial tickets due to inadequate trouble shooting and palming faults off on them instead of dealing with them.

[Maggy]

I don't go with BT because I found them to be totally useless even when they had been privatised.Before that the public service was horrendous too.

If a company cannot get it's basic function of supplying a working landline like BT seemed to have trouble providing to me and then I discover that a Cable company can and will supply me with a rock solid land line that can survive the wildest of weathers I'd be daft to stay with a company that can't.

[Chrysalis]

Again - swings and roundabouts, I've had to wait a week for VM to fix a telephony issue. Phone was totally dead for a week.

All arbitrary and variable depending on where in the country you are and the workload at the time. If VM's service techs are all busy in your area you'll have a long wait too, and the fees are due to the way Openreach work, if VM are forced to wholesale I've no doubt they'll be doing similar to avoid their wholesale customers loading up their techs with trivial tickets due to inadequate trouble shooting and palming faults off on them instead of dealing with them.

well even a long wait is better than nothing, on paper openreach may not be that bad but the real life situation is with many isp's its now not possible to get a openreach engineer at all, the isp's are too scared of the fees. Even if an engineer does come their immediate priority is to look for a fault in the end user equipment, so even then getting line faults fixed is tough.

As I have said on tbb many times, I believe openreach to be breaking trading standard's with their procedures and only get away with it due to their monopoly on phone lines in half the country.

Dont forget you talking to a guy whos had 10 or so engineers out (before openreach started this charging rubbish) and they all failed to fix an ongoing problem on my line, they all failed to check the pole feed and all failed to do a basic hawk test. The only reason I had BT for so long was VM bb was not available here. Now I will likely cancel within a couple of months.

So whilst both companies have variables, VM can be very slow when congested and BT can be very slow with a poor line, the difference is how the two companies approach faults. BT make a lot of money by not properly honouring what should be their obligations to customers, its my view yes, people will disagree. Aaisp have a good article on their thoughts on the fault procedures and they are a well respected isp. They have been forced to tell their customers to hide equipment during visits to force the engineers to diagnose the line instead of end user equipment.

If VM were to wholesale unfortenatly I do see their fault procedures becoming a mess if that were to happen as I suspect ofcom would then start fiddling and telling VM to pull the openreach stunt pretending their techs dont work for them. Given VM's capacity issues as well I hope they are not forced to wholesale.

[Ignitionnet]

Again swings and roundabouts I'm sure there are VM customers who've had multiple engineer visits, service techs blaming their equipment, etc.

Anyway a bit of a divergence from the upstream speeds...

[Chrysalis]

Indeed, will end it here with the agreement in some cases VM is better and in others BT is better. So we can get back to upload discussion.

That article I read from your blog, which indicated a limited frequency range, I assume that means there is a limited capacity for adding new docsis cards, so is it the case its not easy to just add new upstream channels? and do docsis2 need more frequency and snr than docsis1.1?

[Ignitionnet]

Bits of upstream spectrum are unusable due to noise ingress on certain frequencies. Things like hospital pagers, baby monitors and the like operate within the range that cable uses as a return path.

DOCSIS 2 upstreams go up to a width of 6.4MHz which brings its' own challenges, and there are issues over transmit power and SNR that go with it. Obviously to run at 32QAM and 64QAM requires a higher SNR than 16QAM and wider channels both require a cleaner network and have some other more esoteric issues too.

[Chrysalis]

ok understood, so basically noise needs cleaned up a bit prior to such rollouts. Not just a case of swapping out the cards.

[Kymmy]

Just out of interest what is the upstream frequency range??

[Ignitionnet]

Specifications are:

DOCSIS: 5 - 42MHz
J-DOCSIS: 5 - 55MHz
EuroDOCSIS: 5 - 65MHz
DOCSIS 3: 5 - 85MHz.

Individual network areas could be

5 - 30MHz
5 - 42MHz
5 - 50MHz
5 - 65MHz

[Kymmy]

Thank you :clap:

[Ignitionnet]

Not a problem at all. :tu:

[Chrysalis]

is the frequency allocated to tv, flexible. eg. could VM choose to ditch 100 channels and then have a load more capacity for bandwidth?

[Ignitionnet]

is the frequency allocated to tv, flexible. eg. could VM choose to ditch 100 channels and then have a load more capacity for bandwidth?

If you mean could they ditch some downstream and turn it into upstream no, it's not that simple.

Downstream wise they can put anything they want anywhere they want on their network.

[|Kippa|]

What is the maximum upload speed you could theoretically get out of cable. I'm not talking about within the docsis 3 specs I am talking about in general if they pushed the technology to the limits.

[Chrysalis]

If you mean could they ditch some downstream and turn it into upstream no, it's not that simple.

Downstream wise they can put anything they want anywhere they want on their network.

even extra downstream would help as we know some lagacy have congested downstream. But bad news on the upstream tho.

So once the upstream is all used the only thing they can do is a reseg?

[Ignitionnet]

What is the maximum upload speed you could theoretically get out of cable. I'm not talking about within the docsis 3 specs I am talking about in general if they pushed the technology to the limits.

The specifications are the limits of the technology right now.

With those in mind the answer is...

((85-5) / 3.2) * 13.5 = 337.5Mbit/s

---------- Post added at 11:01 ---------- Previous post was at 10:59 ----------

even extra downstream would help as we know some lagacy have congested downstream. But bad news on the upstream tho.

So once the upstream is all used the only thing they can do is a reseg?

Virtually all congestion now is upstream, downstream issues being symptomatic of that.

The answer to the second paragraph is simply it depends.

[Chrysalis]

Virtually all congestion now is upstream, downstream issues being symptomatic of that.

The answer to the second paragraph is simply it depends.

you said my old legacy port was downstream congested. and indeed the min latency went up (as suggested in another post of yours) during peak rather than just the jitter which suggested downstream congestion.

This makes me believe its a bit more common then make believe, I would find it hard to believe I was on the only congested downstream port in the entire VM network.

I was also told on the phone my old legacy setup had 2 downstreams mixed with 4 upstreams instead of 1 mixed with 4 as the single downstream port got completely overwhelmed. This may then go to some way tho in explaining why my old upstream was in such a bad state, as it would typically be 8 upstreams per 2 downstreams? I had 4 qpsk upstreams with 2 downstreams.

[Ignitionnet]

I did say 'virtually all' I didn't say there was only one port. The overwhelming majority of congestion is upstream.

Dual DOCSIS using both downstream ports and 4 upstream ports on an MC28U card isn't uncommon. No area on VM's network has 8 upstreams to a single node. The 2 downstreams and 8 upstreams are usually split into 2 groups of 1-4.

[Sephiroth]

The specifications are the limits of the technology right now.

With those in mind the answer (maximum upload speed from cable) is...

((85-5) / 3.2) * 13.5 = 337.5Mbit/s

......

Igni, someone had to ask. The 13.5 multiplier. I'd have expected you to get the number of channels (you did that) and multiply it up by the QAM related symbols per second * the number of bits per symbol.

So you were working on DOCSIS 1.x (3.2 MHz) and I would have expected 16QAM (4 bits/symbol) * 2.56 Msyms/sec = 10.24 Mbits/s.

So where did 13.5 come from? It doesn't fit 32QAM nor 64QAM either.

[Ignitionnet]

It's the IP throughput of a 3.2 wide 64QAM channel.

[Sephiroth]

It's the IP throughput of a 3.2 wide 64QAM channel.

Gotcha - as distinct from the RAW data rate (15.36 Mbps). You pixie you.

[danlufan]

Well theoretically you can use DWDM on upto 80 channels (potentially even more) with each channel carrying 2.5 Gbps. Sigbritt Löthberg's home was supplied with a 40 Gigabits per second connection as a test in sweeden, so the sky is the limit when it comes to theoretical speeds, however network bandwidth issues will always step in, always has to be processed and monitored somewhere :)

[Sephiroth]

Well theoretically you can use DWDM on upto 80 channels (potentially even more) with each channel carrying 2.5 Gbps. Sigbritt Löthberg's home was supplied with a 40 Gigabits per second connection as a test in sweeden, so the sky is the limit when it comes to theoretical speeds, however network bandwidth issues will always step in, always has to be processed and monitored somewhere :)

So does cost have to be factored in. We all know that when we're dead and gone, UK will be FTTH as will everywhere else in the techno-economic world - but right now competition forces prices down and investment cash for a large scale roll-out is a difficulty.

That said, if VM pilots pole-strung fibre (there's more to it than just saying that), we could see the future sooner.

Anyway, this download may be an interesting read. http://ibcn.intec.ugent.be/te/Members/PhD_KoenCasier.pdf

[Ignitionnet]

Well theoretically you can use DWDM on upto 80 channels (potentially even more) with each channel carrying 2.5 Gbps. Sigbritt Löthberg's home was supplied with a 40 Gigabits per second connection as a test in sweeden, so the sky is the limit when it comes to theoretical speeds, however network bandwidth issues will always step in, always has to be processed and monitored somewhere :)

The question asked was what the theoretical upstream speed of cable was, so it's somewhat irrelevant what DWDM can do.

2.5Gbps on a single wavelength was obsolete a while ago, we're up to 100Gbps on a single wavelength now :)

[danlufan]

The question asked was what the theoretical upstream speed of cable was, so it's somewhat irrelevant what DWDM can do.

2.5Gbps on a single wavelength was obsolete a while ago, we're up to 100Gbps on a single wavelength now :)

I was kind of being pedantic and just putting some pie in the sky statements out there.

I think it would just be nice to see a 5-10mb upload speed, i have noticed the difference in having 1.6mb recently since i had a 50mb line put in,

what sort of timeframes do people estimate VM changing the upload speeds?