30M What do VM do to make this happen?

Efour · 09-09-2011, 01:26

[img]Download Failed (1)[/img]

At Midnight on the dot..... Everything returns to "normal" How is this traffic management actually helping anyone? Is this the cause?

Someone explain it to me...Im perplexed.

Sirius · 09-09-2011, 07:43

Quote:

Originally Posted by Efour

At Midnight on the dot..... Everything returns to "normal" How is this traffic management actually helping anyone? Is this the cause?

Someone explain it to me...Im perplexed.

Its the magic traffic shaping fairies

.

jb66 · 09-09-2011, 08:22

It doesnt look like midnight on the dot to me, more like 2am

kwikbreaks · 09-09-2011, 08:35

The packet loss decreased considerably at midnight though.

I'd say a display when the connection isn't being heavily used would be more likely to show whether or not any blame lies with VM. If it wasn't being used fairly heavily then it looks to me like it must be an oversubscribed area.

weesteev · 09-09-2011, 09:17

Packet loss reduced by latency remained an issue for a further 2 hours, this could well be spikes in overuse in your area. Its basically from 4pm to midnight, suggests high usage, if its causing issues with your connection then I would definitely speak to Tech Support and get then to check local UBR load.

Sephiroth · 09-09-2011, 09:57

Seeing that TBB is an external application sending pings to the OP's router, the packet loss could be on the return/upstream. That might be due to upstream congestion or due to a problem on the OP's circuit (like upstream SNR). We don't have the benefit of the OP's stats and event log to try and join all the evidence together.

Ignitionnet · 09-09-2011, 11:51

It's a load issue, TS have confirmed this.

http://community.virginmedia.com/vir...d=68300#M68300

Scheduled fix next month.

http://community.virginmedia.com/t5/.../558469/page/8

thenry · 09-09-2011, 11:56

brilliant. what will it take for VM to sort this out across their whole network ? Its like waiting for a complete network failure!

Efour · 09-09-2011, 12:13

Hello, Yep VM are saying that they are dealing with it originally this month now next month.

I was wondering why it ended at Midnight (effectively) with such a huge decrease in loss.
It cant be a coincidence? or its 1 or 2 extremely heavy upstream users decide its bed time at 12am?

Why don't VM cap upstream when there are problems like this. would that help at least id be able to play some games.

as for line stats all are normal within range.

Ignitionnet · 09-09-2011, 12:57

It could well be a single person switching their torrents off, especially if 100Mb has been released in your area and you're on the same upstream as a 100Mb torrent fiend.

---------- Post added at 11:57 ---------- Previous post was at 11:55 ----------

Quote:

Originally Posted by thenry

brilliant. what will it take for VM to sort this out across their whole network ? Its like waiting for a complete network failure!

Sort what out?

Putting it into context there are more capacity problems than usual, my area is one of them, but they are not that big a % of the total areas and all are local, it's not a generic network-wide fault but a ton of smaller ones.

thenry · 09-09-2011, 13:07

latency. my post was a bit extreme I apologize

Ignitionnet · 09-09-2011, 13:38

It isn't ideal, true, even areas where everything else works as it should latency or jitter can be a little higher than expected.

VM, nationally, aren't a match for most DSL operators on many metrics that don't rely on raw bandwidth.

thenry · 09-09-2011, 14:02

what would VM have to do to fix latency and is it too expensive?

Ignitionnet · 09-09-2011, 16:26

Quote:

Originally Posted by thenry

what would VM have to do to fix latency and is it too expensive?

Loads of capacity upgrades and yes, for the prices we're paying right now.

qasdfdsaq · 10-09-2011, 01:48

Quote:

Originally Posted by Efour

It cant be a coincidence? or its 1 or 2 extremely heavy upstream users decide its bed time at 12am?

Usage is actually pretty modal like that.

If you look at the Linx traffic charts (https://www.linx.net/pubtools/traffi...tml?stats=week) you can see there's a pretty quick and steep drop off of traffic towards the end of the day.

That includes all business usage as well though. VM (and all consumer ISPs) tend to see more traffic towards later evenings, and if you look at my VM chart during a fault period last month where changes in load are all the more obvious (and on a fatter pipe) you can see the precipitous drop off around 1-2am as well:

While it only takes 1 or 2 people switching off their torrents, a lot of people also do go to sleep at the same time.