Bqm weird early morning result

https://www.thinkbroadband.com/broadband/monitoring/...

Any ideas why this has occurred, not seen it before.

Even worse now..

In reply to a post by ploddo:

https://www.thinkbroadband.com/broadband/monitoring/...

Any ideas why this has occurred, not seen it before.

What service provider / where?

Looks like some bit of kit progressively failing from 05:30 to around 09:30 - then something else has kicked in (notice the step change in minimum latency about 09:30) - a path change.

However that alternative / fix doesn't look to happy with the load - hence the yellow maximum latency spikes through the working hours, with some packet loss, and finally the much higher late night load between 19:30 to 23:30 has killed average and maximum latency and more severe packet loss.

Cloudscape connect. FTTC. They had been monitoring my connection as it was slow for a while but stopped last week. Looks like that was down to the modem getting too hot in the heatwave. There was 41 disconnects at that time too.
This morning it’s looking more normal. 6 disconnects in the last 2 days. We did have a power cut though.

Well that all tallies. The vagaries of FTTC and excessive heat/wet etc environmental factors. The power faults / interruptions won’t be helping matters either.

In reply to a post by Pheasant:

Well that all tallies. The vagaries of FTTC and excessive heat/wet etc environmental factors. The power faults / interruptions won’t be helping matters either.

Unless you can map it against a rain storm not sure 6am loss would be explained by heat.. Would be more middle of the day issues.

I've checked other Cloudscape ones and jitter is quite bad on most but no early morning packetloss.. but the evening bit seems to suffer on all of them albeit to varying degrees

seb

That's a good point the worst of the dropouts seem to be in that early morning period, you'd expect that to be the coolest part of the day.

Unless something was getting increasingly heat-soaked during the day, but then it would be a late afternoon or early evening occurrence, when the heat soak could be at its worst.

Perhaps "heat" was a convenient excuse, there is something definitely awry with that connection.

In reply to a post by Pheasant:

That's a good point the worst of the dropouts seem to be in that early morning period, you'd expect that to be the coolest part of the day.

Unless something was getting increasingly heat-soaked during the day, but then it would be a late afternoon or early evening occurrence, when the heat soak could be at its worst.

Perhaps "heat" was a convenient excuse, there is something definitely awry with that connection.

We have one feature we want to launch that will help see issues better, but also I want to get another out which will troubleshoot more.. and finally I want to give aggregate access to users.. our internal interface for that isn't perfect yet but it's nice to be able to see other graphs to compare.

seb

Sounds good.

Just been reading another thread over at ISPR about some severe packet loss on the Merula ISP network (OFNL) over the period of the match last night. Looks horrific:

https://www.ispreview.co.uk/talk/threads/packet-loss...

So whatever further BQM diagnostic features you have up your sleeves will be welcome I dare say..😅

Edited by Pheasant (Thu 16-Jul-26 20:57:49)

Good Afternoon

I was pointed to this (I'm not on here half as often as I used to be)

There's actually a couple of linked issues causing some of the network issues this week. On Monday we lost a high capacity dark fibre between two London Data Centres and during the morning our second link (different provider) also suffered a stability issue.

While we restored most stability quickly this did reduce our core capacity so coupled with the higher than normal traffic levels this did cause some packet loss - However this mostly showed on OFNL - non OFNL connections did not show this to any significant degree

The graph on Tuesday morning was actually the result of a single LNS /Router seeing some issues we believe from the network flaps on Monday - once the process was restarted this returned to normal

We do closely monitor and ensure our network has the capacity needed

Sorry if you saw the issue - and hopefully it's now back to normal

Richard

Hi Richard

Full respect to you for coming here and explaining, there are not many (if any) ISPs who would do that.

Thanks Richard

For completeness, I just linked your post above, back to the original ISPR thread here that I was referring too.

Do you also provide wholesale / backhaul connectivity for this threads OP’s service provider, Cloudscape Connect?

Good Afternoon

We do offer a range of services for others - for obvious reasons I can't confirm who they are.

Sorry I can't answer that one as clearly as the other details

Richard

Not quite back to normal but getting there. Very much better though.
https://www.thinkbroadband.com/broadband/monitoring/...

In reply to a post by Pheasant:

Do you also provide wholesale / backhaul connectivity for this threads OP’s service provider, Cloudscape Connect?

Public info which may answer your question
https://bgp.tools/rir-owner/uk.cloudscape

😀

For anyone who wants to follow
https://www.thinkbroadband.com/news/fifa-world-cup-s...

Edited by seb (Sun 19-Jul-26 19:24:21)

There have been a few edits to the WC final update text - how widespread was the issue looking across OFNL in the end? Still definitely an OFNL issue or could it be issues at specific ISPs themselves?

In reply to a post by Adrianuk:

There have been a few edits to the WC final update text - how widespread was the issue looking across OFNL in the end? Still definitely an OFNL issue or could it be issues at specific ISPs themselves?

We saw issues on a few Merula routed IPs which were OFNL on the wholesale end. We didn't see them on the OFNL native (routed by them via Cogent) ones for the final so looks like they may have fixed some bits. Not sure where the congestion may be with Merula.

We also did an update with various providers as the issues are more widespread. A few examples on there across providers.