» ISP News » 

UPDATE2 Peak Time Latency Spikes Return to Haunt Virgin Media’s Broadband

Wednesday, May 17th, 2017 (9:09 am) by Mark Jackson (Score 2,282)
Virgin Media 2014 UK Logo

Customers of broadband cable operator Virgin Media are once again reporting a spike in latency during peak times (typically starting at before 7pm and running until around 11pm), which can cause disruption for latency dependent software / services like fast paced online video games and voice/video calls etc.

Latency is a measure of the time (delay in milliseconds [1000ms = 1 second]) that it takes for a packet of data to travel from your computer to a remote server and then back again (ping). The delay is usually measured in milliseconds (e.g. 1000ms = 1 second) and modern fixed line broadband connections will often have an average latency of anything from around 10-40ms.

A lower score (shortest time) is always best for latency, although the times can be affected by various things such as the performance of remote internet servers, connection technology, ISP network congestion, peering / routing problems and the setup of your own home network etc. We should stress that today’s news is separate from the latency bug that is known to exist inside Virgin Media’s Intel Puma 6 powered SuperHub 3 router (here).

At some point all ISPs will suffer network problems that can cause a spike in latency and if the posts on Virgin Media’s Community forum (e.g. here, here and here) and Thinkbroadband are anything to go by then that’s precisely what has been hitting the cable operator this week (during peak times). Sadly it’s an issue that also cropped up at the same ISP in March.

virgin_media_latency_surge

Customers who suffer the issue are likely to see their latency times more than double (the green and blue bits are what you should pay most attention to above), with some going from around 20ms to nearly 80ms and that’s enough to cause a noticeable impact on some online video games and possibly a few other internet services. However other services may not show any noticeable impact.

The suspicion is that this could be the result of a peering problem between Virgin Media’s national network and the London Internet Exchange (LINX), which has caused similar problems in the past and usually results in a few days or weeks of the same until the ISP rebalances their traffic. We have asked Virgin Media and they are currently investigating (expect a comment shortly).

UPDATE 6:39pm

We’ve just received the following comment.

A Virgin Media Spokesperson told ISPreview.co.uk:

“A very small number of Virgin Media customers (less than 1%) may have noticed a deviation in their expected connection performance when using some tools and websites

This is caused by a link failure resulting in the loss of around 0.98% of Virgin Media’s broadband capacity. This will affect these customers at peak times (after 6pm).

Virgin Media is working hard to restore the capacity as quickly as possible. We apologise to the small number of customers affected for any inconvenience.”

UPDATE 18th May 2017

The same pattern was observed again last night and if past problems are anything to go by then we might continue to see this for a few more days or possibly even weeks.

Delicious
Add to Diigo
Leave a Comment
23 Responses
  1. Joshyy

    Had the same thing happen last night, I didn’t notice it though. Only when I checked my TBB monitor.
    https://www.thinkbroadband.com/broadband/monitoring/quality/share/d870e43c24dd5a47d0947d9b8d8ec2d40a91dac2-17-05-2017

  2. Chris P

    shhh

    the resident Virgin supporters won’t like this story one bit.

    Expect cries of fake news imminently,

  3. CarlT

    This is getting pretty tiresome now.

  4. ian

    virgin suck and have for a while, maybe they’re okay for some but for many its a major issue.

    Glad im out, FTTC from BT is actually way more stable. pretty much 80 mbps 24 hours a day.

    Not 300Mbps at 4am from virgin down to 3 Mbps in the evening. Criminal.

  5. Originally a happy VM customer until Dec 2016

    Was good until December 2016.

    From what I can gather. those customers on RIVER DELTA kit (now Motorola) have the issue. Fine on multi thread tests like speedtest.net but on an actual file download from Apple, Microsoft, Intel or the big boys with plenty of bandwidth its like 6-8mbps and WAY below what it should be. Thinkbroadband gives the true results on single thread file downloads, as well as your own eyes when downloading music, movies (legitimate) and ISO of Windows 10 inside builds…..used to be seconds or minutes and now 660KB-3.4MBps on Vivid 200.

    No power or SNR issues here and used a Suprhub2ac as well as a HUB 3.0 – yet it did work before December at 20MBS on a file download in a browser.

    VM Support and networks keep saying “no fault” – and as they want give a fault code, their retentions/leaving department wont process a break in contract, so CISAS next I guess after being a customer since 2008.

    Every time I have changed my package this company locks you into a 12 month contract too. This time they didn’t even tell me or send me a contract too, yet I have a proven history of returning them agreeing previously in the past. They fobbed that off as “not electronic this time by email but sent in post” – err why? It didn’t arrive anyway.

  6. marzencja

    I’m waiting 6th week for sorting out the issue, after replacing hub (to Hub 3.0), it’s a nightmare, I stay without internet since Monday….1% percent of customers???so Virgin media can’t be bothered..money will flow anyway?

  7. CarlT

    That statement is absolute BS.

    That said, if losing 0.98% of capacity degrades customer experience PR people talking out of their hindmost is the least of their issues.

  8. Ultraspeedy

    SO turns out its peering with Linx again. Linx really need to sort there-selves out.

    • CarlT

      LINX are not at fault in any way.

    • Rich

      Confirmed, definitely peering… Speed tests and ping times via many Ookla Speedtest sites are unaffected for me. (I guess they go a different route to TBB monitor). If you also start a VPN which routes differently the issue with TBB graphs goes away. Any sites which were laggy also spring into life. No shock really considering last year a simile (albeit power related one) struck down BT…
      http://www.ispreview.co.uk/index.php/2016/07/bt-network-fault-affecting-uk-internet-broadband-services.html

      Linx were also a major thorn in the later and last years of BE Broadbands life.

      As a user on VM forums also pointed out…
      GO via TBB speedtest linx dependant test and you see things like this…
      https://www.thinkbroadband.com/_assets/speedtest/button/1495048500727851855.png
      go via sppedtest.net…
      http://www.speedtest.net/result/6305290681.png
      both tests less than 2 mins apart.

      Turn on a VPN the bump to TBB goes away, which is no shock as if this was actual congestion the graphs would not just be a sudden high blip it would gradually increase or decrease as the network becomes more and more or less and less congested. Congestion does not just suddenly appear and vanish.

    • Ultraspeedy

      “if this was actual congestion the graphs would not just be a sudden high blip it would gradually increase or decrease as the network becomes more and more or less and less congested. Congestion does not just suddenly appear and vanish.”

      Exactly 🙂

    • CarlT

      Err yes, on core links it does suddenly appear. As soon as the link makes out latency rises as packets sit in buffers for longer waiting to go across the congested links. As the buffers are small relative to the size of the links there’s not much time for packets to wait.

      As the TBB meter uses small packets these get a better experience than larger ones.

    • CarlT

      -makes +maxes

      The issues Be had were also due to them not having enough capacity to connect to LINX. When LINX have issues dozens, if not hundreds, of companies have problems.

      Ethernet links on core and peering networks do not behave in the same manner our broadband connections do.

      VM actually admitted to an issue in the story and elsewhere but, whatever. Let’s blame LINX even though everyone else connecting to it was fine, some of whom will share switches and line cards with VM’s connections to the LANs.

    • CarlT

      Just read the post on TBB. Not sure what they think using a VPN proves beyond that the route to the VPN doesn’t use the impacted link.

      That the problem is peak times only should make it pretty clear it’s congestion.

    • Ultraspeedy

      Routes around the issue by not using Linx. If everything works fine except for routes via Linx than it is obvious were blame lies.

    • CarlT

      Indeed. Given only VM had issues VM’s connections to LINX. LINX are not responsible for how heavily utilised operators allow ports to be and, indeed, VM have been warned more than once in the past for allowing their LINX public peering ports to congest.

      LINX just provide switch ports, it’s up to VM whether they order enough capacity to cope.

      In any event, VM have some upgrade and consolidation work taking place later this year to increase capacity. They are aware of the issue and are taking steps to resolve it.

      Given VM are upgrading their network to fix the issue can’t really blame LINX. The traffic stats show no issues on their side last week. https://www.linx.net/tech-info-help/traffic-stats

    • Ultraspeedy

      IT is not congestion, you can run speedtests on speedtest.net and have no issue at all. 99% of all websites likewise work fine, there is no slowdown because things are not congested. I suggest you look at some recent posts on VM forums.

  9. Darren

    VM are useless and this is a perfect example of why. Thing go wrong and if you have no plan for it customers suffer.

    • Ultraspeedy

      Linx at fault not virgin, but carry on ranting like a loon.

    • CarlT

      Were LINX at fault you’d imagine it would impact the hundreds of other companies connecting via it.

    • Ultraspeedy

      Not really different routing and IP blocks, much in the same way back in 2010 they had issues that only affected those on certain BT based providers and not on Virgin…
      http://www.ispreview.co.uk/story/2010/01/12/uk-broadband-isps-impacted-by-linx-and-bt-congestion-problems.html

    • CarlT

      LINX is not a routed network, it’s a switched one. The operators are all on the same LAN and all have IP addresses in the same subnet.

      VM share line cards and switches on that LAN with other operators.

      For example they have a 10Gb port on switch edge2-rbs PIC 7/0 – they share that PIC with Saudi Telecom Company. They have a port on PIC, edge5-thw 3/0 which they share with EE and one on edge5-thw 3/1, a PIC they also share with EE. They’ve a port on switch52 slot 8, they share that slot with Microsoft and Amazon.

      Failure of ports would be on the LINX service status. LINX service status was clean. VM acknowledge congestion on their network and resolved the issue through manual traffic engineering.

      The two issues you are talking about in the ISP article are entirely separate. One was failed ports at LINX causing congestion between two POPs, the other issues on the BT Wholesale network.

      The BT Wholesale issues were due to a chassis upgrade being late for the capacity required, so they ran out of ports they could put into LAGs. No more physical ports available on the line cards, no slots available for new line cards, hence BRAS POPs congested.

      BT Wholesale’s woes dealing with the increased traffic as 21CN ramped up and FTTC came online are well documented.

      BT Wholesale do not deliver traffic to their CP customers via the LINX public LAN. The specifications of the product are on the BT Wholesale website and require point to point fibre.

      Hope this helps.

  10. Ultraspeedy

    “LINX is not a routed network, it’s a switched one. The operators are all on the same LAN and all have IP addresses in the same subnet.”

    WRONG…
    Linx operates two physically separate networks at 10 different locations. They are not all the same addressing either.

    One network operates on a 195.66.x.x range the other on an Ip of 206.55.x.x

    Further more the switch a group is on means little, its the port that is being used and what is available which which is an issue in cases like this.

    Virgins traffic goes over core 10 IPs, yet linx routes them all over a single port that being 5089.

    They do similar stupidity for Sky who have 5 IPs (another company that has suffered linx peering dumbness on several occasions).

    Its nothing to do with not having enough either…
    Opal Telecom have 20 IPs and linx still routes that all over a singular port of 13285.

    BT on the other hand get preferential treatment it would seem, excluding IPs for their Ireland division they have only 4 more than Opal (24 total) yet get the comfort of having 2 ports used those being 2856 and 5400.

    Linx are a dumb organisation. For one of the banks they have peering arrangements with that being Lloyds Bank plc, they didn’t even do that right or securely until only just over a year ago.

    Lloyds Bank plc have 5 IPs with Linx and until just over a year ago all 5 used port 8435, until i guess someone suggested to them that may not perhaps be the most secure thing to do and one of the 5 IPs was changed to operate on a port of 49758.

    Now you know why all those fake spam bank emails often pretend to be Lloyds LOL.

    In short Linx = cluster (BLEEP).

Leave a Reply

Your email address will not be published. Required fields are marked *

IMPORTANT: Javascript must be enabled to post (most browsers do this automatically). On mobile devices you may need to load the page in 'Desktop' mode to comment.


Comments RSS Feed

* Your comment might NOT appear immediately (the site cache re-syncs periodically) *
* Comments that break our rules, spam, troll or post via fake IP/proxy servers may be blocked *
Promotion
Cheapest Superfast ISPs
  • Vodafone £20.00 (*25.00)
    Up to 38Mbps, Unlimited
    Gift: None
  • Origin Broadband £23.89 (*31.58)
    Up to 38Mbps, Unlimited
    Gift: None
  • Plusnet £24.99 (*33.98)
    Up to 38Mbps, Unlimited
    Gift: None
  • Sky Broadband £25.00 (*38.99)
    Up to 38Mbps, Unlimited
    Gift: None
  • Hyperoptic £26.00 (*35.00)
    Up to 100Mbps, Unlimited
    Gift: None
Prices inc. Line Rental | View All
Poll
*Javascript must be ON to vote*
The Top 20 Category Tags
  1. BT (1970)
  2. Broadband Delivery UK (1340)
  3. FTTP (1291)
  4. FTTC (1255)
  5. Openreach (986)
  6. Politics (976)
  7. Business (877)
  8. Statistics (787)
  9. Fibre Optic (759)
  10. Mobile Broadband (713)
  11. Wireless Internet (646)
  12. Ofcom Regulation (642)
  13. 4G (598)
  14. Virgin Media (596)
  15. FTTH (539)
  16. Sky Broadband (466)
  17. TalkTalk (441)
  18. EE (385)
  19. Security (316)
  20. 3G (276)
New Forum Topics
Helpful ISP Guides and Tips
»
»
»
»
»
»
»
»
»
»
»
»
»
»
»
»
»
»
»
»
»
»
»
»
»
»
»
Promotion

Copyright © 1999 to Present - ISPreview.co.uk - All Rights Reserved - Terms  ,  Privacy and Cookie Policy  ,  Links  ,  Website Rules