packet loss between Plusnet & Virgin Media (reoccurance)

[#1]

This evening the previously discussed issue of packet loss between Plusnet & VM (see http://community.plus.net/forum/index.php?topic=137657.32) has returned with the same symptoms as before. Issues started around 6:40PM 28/06/2015 and are still ongoing as I write this post.
Is anyone from Plusnet able to comment?
Thanks,
Dean

[#2]

There's nothing I'm aware of.
Let me have a word with the relevant people.

[#3]

This isn't a direct Plusnet issue.
Matthew: Your network guys need to ensure they aren't using the LINX Juniper LAN to reach anything advertised by Virgin Media. The affected prefixes are, I think, also advertised on the Extreme LAN so shouldn't be too stressful.

[#4]

I access an email server through VM's network, see attached, the ping plotter results (I have masked the last two hops).
A lot of packet loss, and high latency at this time of the day.

[#5]

Quote from: Ignition

This isn't a direct Plusnet issue. Matthew: Your network guys need to ensure they aren't using the LINX Juniper LAN to reach anything advertised by Virgin Media. The affected prefixes are, I think, also advertised on the Extreme LAN so shouldn't be too stressful.

VM overloaded those peering points ignition?

[#6]

We've raised it to Virgin and they're investigating.
Not much we can do at the moment as it's in their court so to speak.

[#7]

Hmm my BGP may be a bit rusty but I'm pretty sure that you guys control how traffic leaves your network, it's the return path that's the tricky one as it relies on things like the other side honouring MEDs if you set them.
Given this issue is on what would be your forward path there're definitely actions that could be taken. Local preference will take precedence over anything VM have in place bar their not advertising the routes anywhere besides across the Juniper LAN, and transit can be used in that instance.

[#8]

Our office connection is on VM (Big Red 100Mbps). For months now I've had issues during peak periods connecting from home (FTTC) over IPSec L2L (Juniper SRX both ends, managed by me). Ping times to devices behind the remote IPSec end-point are up in the 90-100ms levels, and often I see 30-60% packet loss. Now that could be either Plusnet, VM or both networks suffering congestion, however switch to SSLVPN client, everything is fine. Also pinging the gateway router in-front of the remote SRX shows no congestion.
We have several other sites on L2L into the VM connection, and remotely trying to ping from one of them (Cogent hosted for example) into VM site, the response times are 10-20ms, so on the surface the issue would appear not to be a VM one, or, as ignition mentions, the route to/from could be taking a different, less congested path.
[me=millap]scratches head....now what else could it be.......[/me]
I wonder if the Plusnet traffic shaping policy might be effecting things....
In my professional life, we sell/support QoS appliances (Blue Coat Packetshaper, Allot NetEnforcer), and it's surprisingly, alarmingly easy to break legitimate traffic unexpectedly using these appliances.
I wouldn't rule out local exchange congestion either (as noted elsewhere in these forums). I'm on SSCHI which has (apparently) had issues with capacity, which are (apparently) fixed.
Interesting that it all clears itself up between Plusnet and VM at later at night though 😉
millap

[#9]

Quote from: Ignition

Hmm my BGP may be a bit rusty but I'm pretty sure that you guys control how traffic leaves your network, it's the return path that's the tricky one as it relies on things like the other side honouring MEDs if you set them. Given this issue is on what would be your forward path there're definitely actions that could be taken. Local preference will take precedence over anything VM have in place bar their not advertising the routes anywhere besides across the Juniper LAN, and transit can be used in that instance.

From what I understand of what I've previously been told, the routes PN's outgoing traffic takes are automatically configured based on the shortest path, not the quickest.
They don't like / won't / can't change the path that's been automatically picked, this is why some popular torrent indexing websites will occasionally fall foul of being blocked, because a peer network that is blocking them happens to be the shortest that day.
This may also vary from gateway to gateway, so terminating your PPP session and re-establishing it until you hit a different family of gateways (ptn/pcl/ptw-ag rather than ptn/ptw-bng) may also eliminate the problem in the short term.

[#10]

Ive been having same issue started last night for me I posted in the old thread but some kind soul sent me here
Thanks

[#11]

Just created an account to say I'm seeing this too. Are there any updates from Virgin? Or is there anything that can be done as a workaround in the short term?

[#12]

It seems to have been an ongoing thread on the VM forums for a while, see:
http://community.virginmedia.com/t5/Set-up/Mumble-Server-issues-for-Virgin-Media-users/td-p/2840116/
and
http://community.virginmedia.com/t5/Speed/Strange-anomaly-every-day-at-11pm-and-grows-in-strength-to...

[#13]

VM are not the best company for resolving congestion issues, might be a long wait guys.

[#14]

Dont know if this is directly related but I usually  video Skype a friend on Sunday nights who's on Virgin. Tonight it was unusable. I could see & hear him ok but he said my pic was awful & couldnt hear me at all. This was the same on 3 different devices I tried wired or wireless. My speedtests were fine showing at my max of 38mb down & 20mb up. Streaming full hd from Amazon was also ok. This problem is not uncommon when Skyping that friend so wondering if it's connected with this PlusNet/Virgin problem or its just bad Skype connections.

[#15]

Is there any update on this? As the constant packet loss of 50% and connection drops to my friends Teamspeak server (He is on Virgin Media) is getting a bit beyond.