Broadband News

BBC Watchdog takes a bite out of Virgin Media

Following up on our lengthy item that laid out analysis of how Virgin Media is performing in its different areas of the UK and how we are now poised to keep an eye on their performance more closely (and the other providers and technologies too) we felt it good to add a few more words around what was broadcast on BBC Watchdog on the evening of July 5th.

The issues with over utilisation and core network issues have been many and varied at Virgin Media, and the research gives some idea of the scale of the issue, and while there was a short period of time where we saw very few getting the best speeds things appear to be improving. The key now is whether these improvements are sustained and for those customers who have been affected in some cases for a couple of years how good will Virgin Media be at compensating customers and in the future being more upfront about issues rather than leading customers down the path of having to sit in for multiple engineer visits to fix things that are nothing to do with the final coax drop into the property.

For those affected as highlighted on the programme there is no automatic compensation path, and only by pestering individually will Virgin Media look at each case, in the first instance if the normal moans go unheaded people need to ask to escalate the complaint so that someone more senior in customer support is dealing with you and utilimately if you have the evidence to show prolonged periods with problems the Alternate Dispute Resolution scheme is the way to go, and given this costs providers money they are invariably keen to resolve things rather than go down the slow expensive ADR route.

Feedback from the audience on BBC Watchdog was that while £10/m price reductions were pretty common, when you are still paying £30 to £40 a month for the broadband service and finding you cannot use it as you want this is not that good a deal, there was even one told because they got such a good deal when they signed up no price reductions due to poor speeds would be given. A further comment that echo'd with others was that the what seem like twice a year price rises very quickly erode any of these savings to.

Compensation for faults is something that should eventually become automatic but that is for total loss of service, and while some have been seeing that with Virgin Media those with broadband speed problems would not be automatically compensated based on what we know about the changes on the way, clearly with the importance of broadband to our daily lives this may have to change rapidly.

Looking forward, we hope to be able to report in our analysis of July speed test results that the Virgin Media areas have seen further improvements and we will be keeping a close eye on the the FTTC, various FTTH providers and the emerging G.fast products.

The Virgin Media perhaps highlights a problem with the UK broadband scene, no single provider wants to push the boundaries too hard, but rather it is all about appearing slightly better than their competitors in adverts to gain customers, rather than being world leading in their own right. Virgin Media which is part of the much larger Liberty Global group has other providers already selling vastly faster speeds than are sold in the UK, so we say bring on the DOCSIS 3.1 network with upgrades and show what is possible. 

Comments

My view is that the problem with the UK Broadband Scene is more about pricing. There has been a race to the bottom by BT, Sky & Virgin Media that has resulted in artificially low prices for high speed services.

This creates an environment where Virgin Media has no incentive to push beyond say 300Mb/s, why should they when no one else is offering wide spread available speeds near that?

In Liberty Global countries the speeds available represent the competition in the market - i.e. Romania has 1Gig PON to the MDU from the local incumbent therefore UPC Romania have to compete to grow.

  • Broadthought
  • 21 days ago

@thinkbroadband at least @OpenreachGB has hand back speeds if the connection falls below a certain speed, looking a… https://t.co/RMzkYkQiH7

  • @WorkingtonFibre
  • comment via twitter
  • 21 days ago

I can understand including WiFi as an overall measure of customer experience, but it does introduce a whole raft of variables. Some, like WiFi router quality, the ISP has input into, but others, like local congestion, user equipment, router placement, internal walls and so on are not.

It points to the problem of defining and measuring QoS standards in a real-world situation. If ISPs are to be held to account for performance, it has to be objectively measurable. That's tough; for many years I worked on IT performance in Enterprise environments and those were bad enough to measure.

  • TheEulerID
  • 21 days ago

It was not clear whether wifi speeds were being talked about - they should make sure tests are on a *wired* connection, and say wifi speeds will be much lower... and not much sign of what speedtests they were using...

I know it was dumbed down a lot, but this is VM200/300 they are taking about!

  • comnut
  • 21 days ago

I probably placed this against the wrong story. TBB include WiFi tests in their throughput measures. Ofcom doesn't include WiFi in its throughout measures whilst TBB does.

  • TheEulerID
  • 21 days ago

The speeds mentioned on the show eg during secret shopper section were wired.

Had time to talk to audience and issues were same on ethernet, though some advice was shared around on getting better from the wifi e.g splitting ssid to make it easier to avoid congested 2.4 spectrum

  • andrew
  • thinkbroadband staff
  • 21 days ago

About time... At my old post code NW10 5AN we had 150Mbps with speeds dropping to 5Mbps every evening (multi threaded). Most Sundays we could not even sustain a non-hd netflix stream with quality being stuck at 288p with buffering. All we got was around a tenner off, but given we only paid £23.50 it was almost half price.

Credit where credit is due, they let us leave with barely any hassle for free but we endured it for at least 3 months. I must say it was so much worse than Skys congestion (which I also experienced in 2012ish), on Sky we had 2Mbps peak hours but streaming did function nonHD.

  • ukhardy071
  • 21 days ago

Broadthought and others

this is the problem with utilization issues with the way DOCSIS works , on Openreach when congestion happens the QOS fluff does it work normally and you just get slower speeds (if that ever happens, only seen this happen in exchanges that are not 21CN still, 8mb locked line running at 2mb)

on DOCSIS its like wifi network setup the more people using it gets slower but there is a tipping point when it gets past say 60-70% upstream utilization (not down)it starts to fall apart as it has to resend more failed packets until it gets to the point where it gets to packet loss

  • leexgx
  • 21 days ago

and you start to see problems where pages take longer to load or stall load issues with video streaming been unreliable and stopping

the downstream part on virgin network is norm fine it tends to be upstream (this is customer end) but due to upstream been overloaded at times it affects download to the point its notable (if virgin would apply QOS to its network where its congested to limit p2p damage)

all this is likely going to be non issue due to DOCSIS 3.1 as they did with 3.0 as Virign never fixs the utilization issues directly they just wait until a newer DOCSIS comes out

  • leexgx
  • 21 days ago

Most home users don't need anything more than 50mbps but this would preferably be constant. For 98% of the time my 13mbps adsl line is more than adequate for my requirements.

  • broadband66
  • 20 days ago

@broadband66 The problem you have on VM is that by the time your speeds fall from 200Mbps to 5Mbps, they do this as the network is so congested, often the packet loss shoots up, pings shoot up, errors shoot up, so it is no longer a usable 5Mbps. I sometimes found devices would say I did not have an internet connection. Normally 5Mbps ADSL would play Netflix fine, but in these affected areas doing anything even loading a website can be a pain. On Sundays I could never stream Youtube, Netflix not even lowest quality.

  • ukhardy071
  • 20 days ago

The really nasty thing with DOCSIS is that when it reaches a certain threshold it will cause a lot of problems with the TCP flow, especially the acknowledgement traffic. If that is interrupted it messes things up well and truly.

Andrew has faith in multi-threaded throughput as a primary measure of service and quality (as it's the multi-thread test that he seems to use to generate the index). Personally I prefer single thread measures as I think that maps more readily to the end user experience. A slow torrent is one thing, but an unusable video stream something entirely different.

  • TheEulerID
  • 20 days ago

True, though VM have virtually no issues with upstream capacity at the moment so it's a non-issue, and better scheduling has helped with this - latency starts to degrade at a much higher level of utilisation than with the original DOCSIS 1 kit.

  • CarlThomas
  • 20 days ago

has anybody got experience of the Virgin FTTH product, that only gets converted to DOCSIS when its indie your house?

Was wondering if that is subject to the same congestion problems?

  • Ripley
  • 20 days ago

I haven't an issue with my upstream at the moment it is around 12mb but my downstream has started dropping from 200mb to 5mb frequently. When i contact them all I get told is to ensure everything is plugged in correctly, as if I didn't think of that.

  • Negduke
  • 19 days ago

Leexgx you are completely wrong. Please stop posting misinformation on VM.

Almost all the issues are downstream, not upstream. Upstream congestion affects latency and jitter horribly before it really starts impacting downstream throughput.

VM can, do, and have been upgrading DOCSIS 3.0 networks extensively. The network started on 4 channels per service group, it's now on 24 in most areas. There have been node splits which are intensifying through network rearchitecting.

Upstream usage on broadband networks is dropping relative to downstream due to legitimate streaming services.

  • CarlThomas
  • 19 days ago

Lastly, yes, 3.1 will be in some cases however the rollout will not be overnight and 3.1 is not like 3.0. It is not backwards compatible and customers with 3.0 modems cannot use the extra capacity. The benefit will be from heavy users taking 3.1 tiers and that providing relief to the 3.0 channels. 3.1 equipment will not be sent to everyone just as 3.0 modems weren't when 3.0 was new.

Just FYI if the issues were actually on the upstream it wouldn't help at all. Liberty Global, in common with everyone else to deploy 3.1 to date, will initially use it on downstream only.

  • CarlThomas
  • 19 days ago

Just saw one other thing that merits comment. Upstream utilisation can go considerably above 60-70% without issues now thanks to better scheduling on the CMTS.

Packets are not dropped at the DOCSIS layer at this level of utilisation. DOCSIS upstreams do not handle user data by bursting and hoping there isn't a collision.

Upstream transmissions are delayed, not dropped. A modem might have to wait briefly before transmitting. It's not until extreme levels of utilisation that the delay is so long packets are dropped. DOCSIS works on scheduled data transmission timeslots.

  • CarlThomas
  • 19 days ago

for those ** normal people ** still wondering about wifi, this may help!
http://www.zdnet.com/article/five-ways-to-improve-your-wi-fi-performance/

  • comnut
  • 19 days ago

It would be interesting to know just by how much upstream packets can be delayed. TCP relies on a reasonably consistent stream of Acks arriving within expected times. TCP determines the RTT dynamically and uses this as a basis for calculating the timeout periods. Ack timeouts does disrupt data flow in the opposite direction. Not just retries, but reduced xmit window sizes.

It does look like DOCSIS 3 is much better than the earlier versions, but it would be interesting to see some traces on what's actually happening. Is it simply downstream congestion, or are there some more complex issues.

  • TheEulerID
  • 19 days ago

I should also add that there are other ways that pinch points can cause problems of this sort by delaying acks and disrupting the smooth operation of the TCP flow control protocol.

  • TheEulerID
  • 19 days ago

Not aware of anyone apart from you who wondered about WiFi, comnut. I'm sure there are issues with people having problems with WiFi but obviously these aren't down to such things. One of the first things VM support ask people to do is connect via a wired connection to the Hub.

TheEulerID a bit of jitter shouldn't make a big difference in the grand scheme. Would expect a few milliseconds to not be the difference between the server's congestion window being fine and it filling.

Re: a dynamic timeout, not familiar with that other than the shinier algorithms and RTT-based congestion detection.

  • CarlThomas
  • 19 days ago

It was a good watch but that Mockeridge guy - what a joke. I left VM over a year ago due to utilisation that they didn't fix after over 18 months of broken promises and going to CISAS. For him to say they don't sell / promote in overutilised areas is a bare faced lie. They did then and they absolutely still do. They may be signed up to Ofcom codes of practice, but they are clearly worthless since this company gets away with mis-selling each and every day and I don't see Ofcome getting off their useles behinds or doing anything about it.

  • Syphadeus
  • 19 days ago

like giffgaff :)

  • leexgx
  • 19 days ago

and there are people that are very busy, and dont want to waste 10 to 30 mins to get through to support....

A few friends of mine DO NOT have wired access, due to many reasons..
- the router is behind the TV in the front room, and their computer(s) are elsewhere..
- they have an agreement with their landlord to use wifi
- their computer has been set up by a friend, as they are very much a newbie...

  • comnut
  • 18 days ago

@CarlThomas

The RTT monitoring is part of a dynamic system which uses a smoothed average (the SRTT) to calculate the timeout value RTO and was what I was referring to. That's in contrast to systems which use static timeout systems (which some protocols running over UDP use). Clearly it's unwise to use the same RTO to a destination in London as to one in Sidney from the UK.

The odd few milliseconds will not make a difference, but if there are larger "glitches" it can trip the TCP congestion control which can lead to sawtooth pattern flow behaviour.

  • TheEulerID
  • 17 days ago

Ah yes, got you. It would take quite a heavy disturbance indeed to trigger a back off. If a station is waiting that long for a slot to transmit it is a horrible situation indeed.

Worth remembering it's something harder to measure with DOCSIS due to the use of acknowledgement suppression. What the modem is actually trying to send isn't what your own kit is.

  • CarlThomas
  • 17 days ago

I realise it is fashionable to 'beat-up' BT but apart from their customer services, when working, any contention problems are very rarely seen. I have had one underground line fault in the last 5 years.

On my Infinity 2 service, 74MBits/sec download is the norm with a minimum seen of around 70MBits/sec. There are no problems with Netflix/Amazon 4K.

  • DaveCheltenham
  • 17 days ago

Caught up with the Watchdog section tonight, thought i'd check what Virgin say about our congested area:

'You can get 50mb as your average speed and 47mb during peak times (on the basic 50mb service)'

Might be an idea if your interested in getting Virgin to chat late at night with them as they offered me free installation and a £50 credit off my first bill.

  • Kebabselector
  • 17 days ago

The CEO was definitely evading the truth, I do believe personally these issues are normal policy for VM now, they concentrating investment to cover new areas, and existing areas, are been left to rot effectively.
To call it growing pains I think is insulting to those who have been waiting "years" for a fix.
Also the sales staff may have "some" blame, but VM still accept those orders in the areas, which the sales staff have no control over.

  • chrysalis
  • 12 days ago

I am in area 21 and have speed tests on record going back 3 years since I joined VM. I agree with chrysalis: Liberty Global have been poring money via project Lightning into cabling up new areas and ignoring existing customers for far too long on the basis that it is cheaper to buy off the (relatively) few complainers rather than make the necessary investment to keep the service up at an acceptable level. LG are not in the business of "predict and provide", much more "do nothing until there is a significant drop in revenue" which hopefully will be masked by new revenue from new signings.

  • Mikejc
  • 11 days ago

I have had an ongoing complaint with M/s Karen Ingram, VM Customer Service Director, since early April when speeds dropped below 10% of the 50Mbps that I signed up for over a 4 day period. There were repeated promises that it would be fixed, latest date given was 7 July, and in the meantime VM have kept me sweet with compensation deals and increasing my speed to 70 and most recently to 100Mbps without increasing the monthly charge. The irony is that although speeds have now improved I am still not getting what I am now contracted for. However 85Mbps is a lot better than under 5Mbps.

  • Mikejc
  • 11 days ago

Post a comment

Login Register