50M : Think Broadband Ping Monitor Results (POST YOURS)

[Chrysalis]

Originally Posted by qasdfdsaq

Looks like his problems are all at his local node. Still no proof this bump actually affected any real traffic.

bear in mind I did that tracert after the bump ended for your post request.

I didnt do one during the bump.

I do know all the previous bumps had negative affect on my services tho. I remember you saying last time as well you seem to mysteriously think it only affected tbb and you was wrong.

Do you think VM dedicate a port just for tbb traffic?

and yes the latency spikes craig :0 not unusual for my connection I guess. But I just put my modem on a better US now, I think that channel not as good as the other one.

[qasdfdsaq]

Originally Posted by Chrysalis

I do know all the previous bumps had negative affect on my services tho. I remember you saying last time as well you seem to mysteriously think it only affected tbb and you was wrong.

The previous bump was a problem with their capacity on paths through LINX and could be clearly seen affecting many different routes to VM via LINX.

The current bump didn't affect any of the 5 LINX paths I tested, or two private peering paths. That largely rules out a generic problem through LINX.

That leaves two possibilities, a problem specific to TBB or a problem affecting VM's network in general. Since zero out of 7 paths into VM were affected, that rules out a generic VM problem. That leaves one possibility.

I'm yet to see any evidence it affected anything other than TBB, so if you have anything sensible to suggest to the contrary, please do.

Originally Posted by Chrysalis

Do you think VM dedicate a port just for tbb traffic?

What evidence do you have that the fault lies with a port on VM?

---------- Post added at 09:41 ---------- Previous post was at 09:32 ----------

Originally Posted by Chrysalis

bear in mind I did that tracert after the bump ended for your post request.

Seeing as the ping monitors go through LINX and your Tesco traffic does not, unless there's a generic problem with VM's core network (which there isn't) problems with the two are completely unrelated.

[Chrysalis]

qas I dont know the nature of your testing, you seem to be saying you somehow are able to test specific VM ports. Also that VMs core network has no problems, what exclusive access do you have to determine that?

What I can tell you is the following.

VM confirmed themselves that at least one port was overloaded in the previous instance, they had overloaded it with atroto traffic, when they rerouted that traffic temporarily graphs went back to normal, yet your testing determined there was no such overload. So maybe VM were wrong also?

I have done my own testing using speedtests, as a port overload can slow down speeds without affecting pings, I can affect speeds by rerouting my own traffic and then looking at routes taken for that traffic to determine where congestion is, yes i cannot determine the exact point but I also test with multiple isp's which gives me a reasonable guess.

I have multiple times in the past seen what looks like congestion on VMs core network, the CEO office themselves have even told me of core network issues near my area that they have said have caused me problems. Granted they could have been fobbing me off.

I also have contacts with several isp's they can inform me if their peering point with VM is saturated or not, I have these contacts as I do work for them or have done work for them in the past.

eg. I have had traffic moved from AMS-IX peering point to telia transit and the speed quadrupled (not maxed out tho). The src isp however doesnt have issues sending over AMS-IX to other uk isp's.

Do you also believe there to be no issues in manchester VM connectivity, craig himself has even picked up the performance issues in that area.

Generally when the shared pipe is large eg. 10gigabit, then issues wont show on ping data until its pretty much saturated, it can be at 99% and you will see very little in terms of ping spikes. Its not like a UBR port where it can spike at say 60%.

Now I am trying to understand what you are saying, are you saying you think the local network to tbb has its own congestion, yet that congestion just happens to be affecting VM alone, maybe tbb are sabotaging VM? is that in your thoughts?

[qasdfdsaq]

This is exactly the kind of idiotic far fetched speculation that's ******** me off these days.

You have slow access to Tesco. Other people have a bump on their ping charts. They have nothing to do with each other.

You see one bump now. You saw another bump a few weeks ago. Again, they have nothing to do with each other.

You complain of congestion on VM's core network yet somehow nobody else has seen it and all you can demonstrate is congestion in your local area. They have nothing to do with each other.

The ping bumps everyone is seeing has nothing to do with you, your Tesco shopping speeds, or anything in your area.

Basic network troubleshooting 101:


What is the problem?
High latency

What is it affecting?
Traffic from A to B

Is it affecting anything else going to or from A?
No

Is it affecting anything else going to or from B?
No

Is it affecting anything else going through a common midpoint between A and B?
No

Conclusion:
The problem is localized and probably specific to traffic between A and B.


This ping spike has nothing to do with Speedtest, nothing to do with AMS-IX, nothing to do with previous problems with Atroto, nothing to do with Manchester core and nothing to do with link speed. I don't give a rats arse if pigs are chewing up the cables in Amsterdam or sabotaging VM, they have nothing to do with the issue being discussed. Quit clutching at straws and stop banging on about irrelevant stuff that has nothing to the issue at hand.

Oh and some basic logic on the stuff you've come up with:

You say previously VM confirmed there was a port overloaded due to Atroto traffic.

Are any the symptoms the same as the previous incident with Atroto?
No.

Conclusion: It is not the same problem

You say something about a port being overloaded in a way that affects speedtests but not pings.

Does the problem affect pings?
Yes

Conclusion: It is not the same problem

You say you've seen multiple times congestion on VM's core network.

Is the problem affecting anything else on VM's core network?
No

Conclusion: The problem is not with VM's core network.

You say you have contacts with other ISPs that can inform you if their peering point is saturated or not.

Is the problem affecting anything going through a peering point with another ISP?
No

Conclusion: That has nothing to do with the problem.

You say moving traffic from AMS-IX in the past has caused your speed to quadruple in the past.

Is the problem anything to do with your speed?
No

Does the problem affect anything going through AMX-IX in any way shape or form?
No

Does the problem solely affect traffic not going through AMS-IX?
Yes

Conclusion: That has nothing to do with the problem.

You say something about Manchester connectivity and performance issues there.

Does the problem affect any other traffic going to, from or through Manchester?
No

Does the problem affect any traffic not going through Manchester?
Yes

Is there any difference in the problem's effects between traffic going through and not going through Manchester?
No

Conclusion: That has nothing to do with the problem.

You say something about large shared pipes not showing issues on ping data until its pretty much saturated, and it could be running at 99% without you seeing any ping spikes.

Is the problem a ping spike?
Yes

Conclusion: That has nothing to do with the problem.

[Chrysalis]

I could argue you are spouting rubbish as well.

You post it is not VM side, with nothing to back that up other than your statement. So we will have to agree to disagree.

Just remember you calling my green hills a localised problem doesnt have much clout when we have several other regions with the same green hills on their graphs.

I also sense a sign of disrespect when you consider an issue thats bad enough to cause base latency to sky rocket to not be performance affecting.

Also that I never once said tesco was caused by this spike, I simply posted it was extremely slow on my VM connection (not on my 3G), also not on 2 other connections I tested. I also stated the tracert was from after the time period in question and that the latency spikes on that were not unusual for my area. You suddenly go on the defensive over VM's problems as if you have some stake in their network or something.

Thank you for confirming you disagree with VMs statement by the way, as VM did say it was atroto related. They specifically addressed this in relation to the tbb graphs people were posting.

So several websites/services have issues only with VM but its not VMs fault

[iateallthepies]

Oh my days just to add (among the squabbling) i`m suffering this bump too again.. I`m on Craigs pingotron CMTS is Great Yarmouth..

[Qtx]

Chrysalis, don't feed the tro...

[Chrysalis]

yeah sorry for the ranting will go back to just discussing graphs.

[qasdfdsaq]

Originally Posted by Chrysalis

I could argue you are spouting rubbish as well.

Umm...

You post it is not VM side, with nothing to back that up other than your statement.

Where did I say that? You appear to be spouting made up rubbish.

Just remember you calling my green hills a localised problem doesnt have much clout when we have several other regions with the same green hills on their graphs.

Where did I say that? You just made up more rubbish

I also sense a sign of disrespect when you consider an issue thats bad enough to cause base latency to sky rocket to not be performance affecting.

Neither you or anybody else has presented any evidence that the performance of anything was affected other than the TBB ping monitor itself.

Thank you for confirming you disagree with VMs statement by the way, as VM did say it was atroto related.

Where did I say that? You're making up more rubbish again.

They specifically addressed this in relation to the tbb graphs people were posting.

You said for the last problem several weeks ago. Have they done this in relation to yesterday's graphs?

So several websites/services have issues only with VM but its not VMs fault

Several? Asking for the third time, what other than TBB is affected?

[roughbeast]

Just sticking to what I have observed only: During the evening blip over the last three nights my down speed has been reduced to <50Mb and there has been buffering on streamed HD, something that never happens when there is not a blip.

Whatever its cause this aberration is unacceptable especially considering this is a repeat of symptoms I had less than a month ago.

As before, the blip has a very sharp onset and decline. Something changes abruptly at these times. Here I will allow myself to speculate.

This kind of change is not typical of the gradual build up and decrease of congestion you would expect with folk getting home from work and, later, going to bed. Some change is being made either automatically or manually to network configuration or network load in my area at these times.

Is this local or general? I do not know. How many others are getting this effect and where are they?

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

[iateallthepies]

Originally Posted by roughbeast

Is this local or general? I do not know. How many others are getting this effect and where are they?

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

Exactly the same. I had it between 4pm and 6pm yesterday..

[qasdfdsaq]

Originally Posted by roughbeast

Is this local or general? I do not know. How many others are getting this effect and where are they?

It's affecting every single VM monitor I've seen but has not been shown to affect any other traffic to/from VM.

The onset isn't unlike something caused by congestion, the decline however indicates a sudden switch, e.g. routing got fixed or degraded link brought back online.

The sharpness of these things really doesn't say much, it's the relative sharpness of the increase and decline, which if caused by congestion should be similar at both ends.

[Sirius]

Originally Posted by joglynne

It's back in Manchester.

Had the same last night. I am now at the point of making a formal complaint to the top. I have had ENOUGH of it.

[roughbeast]

Originally Posted by qasdfdsaq

It's affecting every single VM monitor I've seen but has not been shown to affect any other traffic to/from VM.

The onset isn't unlike something caused by congestion, the decline however indicates a sudden switch, e.g. routing got fixed or degraded link brought back online.

The sharpness of these things really doesn't say much, it's the relative sharpness of the increase and decline, which if caused by congestion should be similar at both ends.

I'm not yet convinced by your view on the sharpness of the onset. On my TBB graph there was a normal profile until something like 7.50 pm. By 8.15 pm the blip was at its peak. Admittedly the decline at the other end of the blip was virtually instant, but I would have expected a much more gradual onset if collective human behaviour is the only cause. ie thousands of users caning their connections soon after they get home from work somewhere around 6.00 pm.

[gumpy321]

Mine was the same last night too: