Broadband News

We look deeper into the performance variations Virgin Media users see

Virgin Media has been going through something of a love hate relationship with its customers with what seems to be an increasing number turning to social media to complain about issues with the services ranging from poor TV quality from the cable TV aspect, to streamed IP TV services being impossible to watch for some and all at the time when the ultrafast aspects of the service and the ability to host a house full of party guests using the Wi-Fi are promoted to potential new customers and also being seen by those who are having problems.

So why the focus on Virgin Media? Put simply they have had the most variation in the various metrics we are able to track, and we must highlight that there is evidence suggesting things have improved in the second quarter compared to the first, things are still some way behind competing services. The level of complaints about poor speeds and problems with video streaming extends beyond our speed tester as twitter and facebook shows plenty of people complaining about poor speeds including some where the payload is served from the Virgin Media network.

The quality metric is based on how stable the download speed is during the multiple thread test, or put in simple terms after allowing a short time for a test to start how much does the speed vary, and the mathematics ensure that a totally stable 18 Mbps connection will score the same as a totally stable 73 Mbps connection. For those thinking this is an artificial system to measure quality, consider whether a broadband connection that cannot reliably sustain a speed for several seconds is a good or bad one, as a general rule a poor score in the 2 and above region means we expect people to be seeing other issues. Of course Wi-Fi issues can impact the results, but when providers heavily promote the Wi-Fi capabilities of their broadband hardware it seems perfectly reasonable to include it and additionally increasingly people don't have any devices connected via Ethernet in the home (or even own consumer electronics with an Ethernet port as standard).

Chart of Virgin Media areas with quality of speed test results
Click image for larger version

Median Speed Test Quality for Q2/2017 by Virgin Media Area

The quality metric needs its scale explaining so a rough guide is:

  • 1 to 1.2: Very good connection for streaming and gaming with stable speeds
  • 1.3 to 1.8: Connections are good but there may be the odd rare stutter in downloads or gaming
  • 1.8 to 2.2: Likely to be using Wi-Fi or on a good mobile service. The odd dip in speeds and stutters are noticeable
  • 2.2 to 3: Your connection or Wi-Fi is affecting your experience and web pages may arrive in fits and starts
  • 3 to 4: Connection is likely to be heavily congested and streaming will be difficult. It is worth checking no-one in the property is also heavily using the connection
  • Over 4: Time to find a better provider or if on Wi-Fi use an Ethernet cable

The difference between Q2/2017 and Q1/2017 is so big that only one Virgin Media area made it into the normal range of 1 to 1.5 in the first quarter of the year, where as in Q2 99 out of the 222 areas made it into what we consider the normal range. So while things are considerably better now, they are still far from ideal, and to help people understand the comparisons our local statistics site has a comparison map for the four main broadband technologies with areas plotted by postcode letter group (i.e. GU, HA etc). The change has largely been down to what appears to be significant improvements in service during June as April and May were largely mirroring the results of the first quarter.

Results for one Virgin Media in June 2017
Click image for larger version

Example for Virgin Media area results in June 2017 - Warwick

On visiting our technology comparison map those on a Virgin Media consumer broadband connection will also be shown some additional detail. The area is identified based on the IP address when you visit the site, so those not on cable broadband connection will not see the detail, and we are showing the monthly results so that if improvements continue it will be visible sooner than if we just stuck to the quarterly view that all our mapping uses, or put another way if the Dover area shows a major improvement the monthly figures will show this first.

The worst 10 and best 10 Virgin Media areas for Q2/2017 are in the two following tables along with a variety of speed metrics. The ordering is done by the quality score since using download or upload speed is often skewed by varying mixtures of the popularity of the speeds purchased in an area and Virgin Media has some areas where the top speed available is still in theory 200 Mbps.

The 10 Worst Virgin Media Areas for Stable Speeds in Q2/2017
ProviderQualityDirectionSpeed of bottom 20%MedianMeanSpeed of Top 20%
Southport3.06Down 17.7 Mbps 25 Mbps 32 Mbps 46.5 Mbps
    Up 3 Mbps 5.8 Mbps 6.7 Mbps 10.7 Mbps
Ashton-under-Lyme2.99Down 17.4 Mbps 27.3 Mbps 30.4 Mbps 43.2 Mbps
    Up 3.1 Mbps 5.8 Mbps 6.7 Mbps 10.7 Mbps
Dover2.83Down 20.8 Mbps 31.3 Mbps 35.3 Mbps 35.3 Mbps
    Up 4.4 Mbps 7.1 Mbps 8.3 Mbps 11.8 Mbps
Lisburn2.71Down 20.8 Mbps 31.3 Mbps 35.3 Mbps 49.8 Mbps
    Up 3.2 Mbps 6.6 Mbps 7.2 Mbps 11.3 Mbps
Macclesfield2.65Down 17.9 Mbps 29.1 Mbps 32.6 Mbps 49.9 Mbps
    Up 3.1 Mbps 5.7 Mbps 6.3 Mbps 9.7 Mbps
Blackburn2.64Down 14.3 Mbps 22.1 Mbps 24.8 Mbps 33.1 Mbps
    Up 2.9 Mbps 5.3 Mbps 6 Mbps 8.8 Mbps
Barry2.59Down 17.5 Mbps 33.5 Mbps 36.2 Mbps 48 Mbps
    Up 3.6 Mbps 6 Mbps 6.7 Mbps 10.9 Mbps
Folkestone2.58Down 21.1 Mbps 29.5 Mbps 30.8 Mbps 40.5 Mbps
    Up 4.3 Mbps 6.1 Mbps 7.2 Mbps 11.6 Mbps
Stretford2.58Down 11.2 Mbps 23.9 Mbps 27.8 Mbps 40.9 Mbps
    Up 2.7 Mbps 5.5 Mbps 5.6 Mbps 9 Mbps
Amersham2.5Down 25.7 Mbps 39.6 Mbps 47.4 Mbps 66.3 Mbps
    Up 3.2 Mbps 6.2 Mbps 7.2 Mbps 11.4 Mbps
The 10 Best Virgin Media Areas for Stable Speeds in Q2/2017
ProviderQualityDirectionSpeed of bottom 20%MedianMeanSpeed of Top 20%
Poplar1.01Down 25.5 Mbps 55.5 Mbps 61.3 Mbps 93.9 Mbps
    Up 2.8 Mbps 6 Mbps 7.3 Mbps 11.8 Mbps
Gowerton1.09Down 28.6 Mbps 43.1 Mbps 52.3 Mbps 66.2 Mbps
    Up 3.4 Mbps 6.1 Mbps 7.2 Mbps 11.6 Mbps
Bromsgrove1.21Down 30.3 Mbps 50.5 Mbps 65 Mbps 99.9 Mbps
    Up 3.1 Mbps 5.8 Mbps 6.7 Mbps 10.5 Mbps
Luton1.25Down 28.8 Mbps 51.9 Mbps 59.9 Mbps 87.5 Mbps
    Up 3.6 Mbps 6 Mbps 7.3 Mbps 11 Mbps
Halifax1.26Down 32 Mbps 52.7 Mbps 62.9 Mbps 93.2 Mbps
    Up 3.6 Mbps 6 Mbps 7.3 Mbps 11 Mbps
Doncaster1.26Down 30.5 Mbps 49.1 Mbps 63.4 Mbps 95.2 Mbps
    Up 3.1 Mbps 5.9 Mbps 7 Mbps 11.5 Mbps
Croydon1.26Down 22.9 Mbps 49.6 Mbps 60.4 Mbps 92.7 Mbps
    Up 3.1 Mbps 5.8 Mbps 6.9 Mbps 11.4 Mbps
Salford1.27Down 31.9 Mbps 54.6 Mbps 69.2 Mbps 104.7 Mbps
    Up 3.2 Mbps 6.1 Mbps 7.2 Mbps 11.3 Mbps
Shepshed1.27Down 25.5 Mbps 49.4 Mbps 60.6 Mbps 83.9 Mbps
    Up 4 Mbps 5.9 Mbps 6.9 Mbps 11 Mbps
Tilbury1.28Down 28.6 Mbps 53.9 Mbps 69.9 Mbps 107.8 Mbps
    Up 3.9 Mbps 6.1 Mbps 7.6 Mbps 11.7 Mbps

One aspect of the Virgin Media network is that you can have individual nodes in the network that serve maybe 500 to 2,000 customers suffering from congestion issues, so in addition to highlighting that we are listing the median figures. For some areas it is possible for a segment of customers to be performing a lot worse than the median. The last few months appear to have taken the node level issues and compounded them with a variety of network capacity issues, and with the all the expansion work under way it may be that engineering staff are being stretched with work connecting new areas and ensuring that node splits and hardware upgrades keep pace with what the public is doing with their broadband connections. Major network expansion while also serving the needs of existing customers is a difficult game, one only has to revisit the issues Openreach had over poor quality work from contractors and missed appointments a couple of years ago at the peak of the FTTC roll-outs to see this is not a unique Virgin Media issue.

Virgin Media generally beats VDSL2/FTTC hands down for download speeds, but we have seen people moving away and while they report yes their maximum speed is a lot lower they are now able to do Netflix and chill evenings without any buffering or gaming nights are not filled with frustration over how their broadband is impacting on their enjoyment.

While we know you can click away on our broadband map that compares the Q2 results for ADSL2+, FTTC, cable and FTTP we thought showing the FTTC versus Cable results might help to clarify, note the scales are identical in all these images.

Map of postcode areas with quality of cable broadband speed test results plotted
Click image for larger version

Postcode Area Level Map of Median Speed Test Quality for Q2/2017 cable broadband connections
White areas are where we see no cable broadband speed tests or sample size is too small to display

Map of postcode areas with quality of FTTC/VDSL2 speed test results plotted
Click image for larger version

Postcode Area Level Map of Median Speed Test Quality for Q2/2017 FTTC/VDSL2 connections

Chart of Postcode Areas and quality of FTTC/VDSL2 speed test results
Click image for larger version

Median Speed Test Quality for Q2/2017 in postcode areas for Openreach VDSL2/FTTC Speed Tests

Comments

Oh thanks TB and Watchdog, my VM speed tests were usually OK - until the programme started tonight and now they have really dropped.
A month ago: https://www.thinkbroadband.com/speedtest/1497015189333086655
Today: https://www.thinkbroadband.com/speedtest/1499284008267413555
I leave VM on Sept 2nd after 21 years. I don't give a hoot how many times they call with offers when I give them notice, I'm having EE installed on Jul 28th :-)
VM has gone really downhill in all respects. I may live in the 6th best area, but not when it comes to (lack of) customer service or value for money.
As for buffe

  • tmcr
  • 10 months ago

An illuminating analysis, even if it's what many have suspected. What would be interesting to know is how many of these issues are caused by inadequate backhaul and how many are due to overloads in the local segments. The former is surely easier and less costly to deal with than the latter as it would require a lot of local changes in local segments.

Of course, GPON technology has got similar potential issues to DOCSIS, but the bar is probably considerably higher and the segments generally smaller.

  • TheEulerID
  • 10 months ago

I thought Virgin offered 200Mbps or even 300Mbps in places? Their agents quoted varying minima on the watchdog investigation, but the lowest of those was 140Mbps. How can the figures from Virgin fit even remotely with your results for the BEST - the speed of the top 20%?

  • Fellwalker
  • 10 months ago

While 300 Mbps is offered in some places and that list is growing take-up and people choosing to buy it is still small.

The top 10/worst 10 are the overall figures, so includes those on old 30 Mbps tiers, 50/70/100/152/200/300/350 so its a real mixture. In our main monthly ISP ratings we do split the core product options out, but this dilutes the data for comparison on area. Additionally we are including speed tests over Wi-Fi, which Ofcom and ISP adverts exclude, but as so many use WiFi all the time the comparison is important.

  • andrew
  • thinkbroadband staff
  • 10 months ago

How are the new full fibre areas performing? VM are currently rolling out proper fibre to our area (Thanet) and it is tempting, but there is always these congestion issues at the back of my mind, so it would be useful to know how these new roll outs are performing.

  • R0NSKI
  • 10 months ago

I see the quality test is made on the multi-thread throughput. I would have thought that a quality test on the single thread is much more telling as that's more comparable to the most disruptive effects like video buffering. Of course the multi-threaded test puts the most strain on an individual link, but if I think the maximum throughput single thread performance is often the most relevant.

Personally I always look for the ratio between single-thread and multi-thread as the vast majority of things I do have one dominant thread. A house full of teenagers might be different though.

  • TheEulerID
  • 10 months ago

Single thread quality may be a better early indicator of issues but also has more variables so would be disputed by more

  • andrew
  • thinkbroadband staff
  • 10 months ago

Really? In what sense are single thread issues prone to more variables? Yes, single thread is more sensitive to overloads as multi-threading to a large extent cheats the TCP flow control mechanisms (which is why it's really bad news if you are trying to stream a film and somebody else is swamping a shared element with a massive multi-stream torrent). But that's precisely just the sort of thing that causes the most service issues as it's generally single-thread that dominates on interactive use.

So I would say being more sensitive is not the same thing as being subject to more variables.

  • TheEulerID
  • 10 months ago

The issues around wireless and tcp stacks and browser behaviour impact more.

  • andrew
  • thinkbroadband staff
  • 10 months ago

@andrew they do, but that's not another factor, it just makes it more sensitive. It just means that multi-threading bulldozes its way through at the expense of everything else. The reality is that the service effecting issues seen by most people are the single threading problems and I think the more sensitive and relevant measure to most people's Internet experience is the single thread one.

This single vs multi-thread issue came up throughout most of my career in IT, and in many more contexts than comms. It's usually single-thread that affects user experiences.

  • TheEulerID
  • 10 months ago

An analogous issue happened when Sun introduced the hyper-threaded T series servers with 8 threads per core. They were designed to maximise throughput, but at the expense of single-thread speed. A high-level (idiot) manager did a deal with a salesman without a clue that their use in many scenarios was disastrous to end user experiences. Response times tripled, even though they coped with total throughput.

Leading with multi-threaded throughput tests considerable understates the impact on end user service in my experience.

  • TheEulerID
  • 10 months ago

VM salesman called a couple of days ago and when my Sky contract runs out in a couple of months he is going to let me have 50 Mps including anytime phone calls for £26 month and one off £20 connection 12 month contract.

  • David-Park
  • 9 months ago

David-Park push hard you can get a better deal than that with em

  • dragonlord6
  • 9 months ago

Is the TBB single threaded test accurate? It never gives me more than 120Mb, but today alone I've done two single-threaded downloads, both over HTTP, at 38MB/sec on my 300Mb connection. One was from my OVH dedicated server in France, and the other was a Vimeo video. This to me suggests my connection is fine and the TBB speed test is at fault.

  • _mike_
  • 9 months ago

Post a comment

Login Register