How is "beyond" determined?

[mikeshand]

I'm getting outages reported as beyond my ISP, and yet the hop count is only 2. How does the agent determine that it is beyond? Is it by comparing the whois of the router with the ISP name? There are broadly similar, but not identical.

Also, when I set up the agent, I got the service provider name wrong, although it now seems to have figured out the correct name itself.

Provider TISCALI-UK TalkTalk Communications Limited, GB

The failing hop (at 2 hops) is listed as Talk Talk communications Limited

Just a bit puzzling.

[OutagesIO_Support]

Hi and thanks for your question.

I'm getting outages reported as beyond my ISP, and yet the hop count is only 2.

I've not looked at your reports but my guess is there are a fair number of private IP's in the mix.

How does the agent determine that it is beyond? Is it by comparing the whois of the router with the ISP name?

There are broadly similar, but not identical.

It basically looks for gateways and public IP's. I'll try to better explain below.

Unfortunately, sometimes, software simply cannot figure something out and it takes a human being to take a closer look.

If you can share your agent ID in this case, I can take a look at have a better idea of what we are talking about.

Typically, if the agent is not able to tell this information, it's because there are a number of private IP's being used along more than one gateway. The agent cannot look up private IP's because anyone can use them so it cannot know when/who the provider is until it gets to at least one public IP.

For example, if you have two routers on your LAN, those are connected to another router, then that is connected to your providers router/modem and they are using public IP's from their router then into their network for another hop or two, the only way to know the setup is by having a human take a look.

We have considered adding a way for members to manually tell their reports which hop is which for cases like this but have not found a way of offering it in a way that would not cause more problems than not.

Also, when I set up the agent, I got the service provider name wrong, although it now seems to have figured out the correct name itself.

Provider TISCALI-UK TalkTalk Communications Limited, GB

The failing hop (at 2 hops) is listed as Talk Talk communications Limited

I think I know what the answer is but I'd like to reply once I can look at your reports if that's ok. Otherwise, I'm really just guessing.

Note that 'beyond' are supposed to mean beyond your provider. Those are more informational than useful because you could ask your provider for help but if most of the problems are with Level 3 for example, most consumers will have little luck getting anyone to respond by an upper level tier.

One thing I should mention is that the algorithms are constantly under development. We've gotten it to work pretty well but it can always be improved. By looking at situations like this one, we can learn and do a little more fine tuning to try and get it to do better. There are situations however where the network is not as 'standard' as most Internet consumers have things set up and it has a difficult time figuring those out without some human intervention.

Waiting for your reply.

[mikeshand]

Thanks.

ID is 126884

Mike

[mikeshand]

In cases it helps. I ran a tracert from our internal network to 62.24.254.202

which gave me

Hops. IP

1. 79.78.1.216 (I.e the wan IP address of our bds router)

2. 62.24.254.202

This seems to confirm that the fault lies between our router and the first hop upstream. I.e. 2 hops as reported by the agent.

I find it hard to believe that this outside the ISP

Looking forward to your analysis.

Thanks.

[OutagesIO_Support]

In terms of the provider information, here is what is going on.

The actual Provider is TISCALI-UK who is also TalkTalk Communications Limited, GB because of a merger.

Your provider is in fact British Telecom which is using/reselling the above network.

Your business service is called Btopenworld which is in fact a service offered by British Telecom. You had this name in your provider question but it is actually the product/service and not the provider. It's not an issue of course but just something we noticed while going over all this.

Now to address the hops classification, can we get a little more information from you.

The second hop is definitely not beyond your ISP and this is the part we need to address next.

There is a mistake in our code from what we can tell or an improvement that we need to make for this kind of situation.

How is this service being delivered, is it wireless, cable, fiber? Do you have a router, a modem, a router/modem, something else?

Basically just trying to get a visual/layout of your setup, topology.

[OutagesIO_Support]

Hi,

Yes, we've found an error in the classification algorithm and are looking at it right now.

Give us a little while to figure this out, test and get back to you.

[mikeshand]

It’s fibre to the cabinet, then copper into the premises ( at least a couple of hundred metres) connecting into a vdsl router. The agent is running in a Linux pc attached by cat 5 to the router (via an ethernet switch).

Thanks.

[OutagesIO_Support]

Here is the update and what we have found.

First, there was a small error in what we call our classification code. It's not really an error, it's more that code is not able to see things in the same way that a human being can and this is part of the ongoing development work at our end, to try and keep refining these things.

What seems to be happening is that British Telecom (BT) is assigning you an IP from its IP pool/s while the actual provider which is TISCALI-UK (who seems to have merged with TalkTalk Communications) is routing using its own IP pool/s. All of the outgoing IP's are owned by the two merged companies so there is no way to tell which are provider or beyond.

Not unusual, just that software cannot know this, a human being has to step in to know this.

There is no way for the software to know these things and no way to really tell it how to know this. However, the one thing we've come up with is that we will try to add a little notice when this kind of situation comes up so that the member can know to look into it and understand what is what in terms of their hops.

We fixed a small error so that the next time your hop two has a outage or problem, you'll notice it will say this is your provider. The next couple of hops however, maybe show up as beyond. In this case, you'll be able to tell/know that your second hop at least is definitely your provider while the next 3 (I think it was) are in fact the upper tier provider.

In testing, we saw that other members who are using BT are showing the same issues so it's up to a human being to make sense of what the software cannot always discover on its own.

I'm not trying to make excuses because we want our service to work as best as it possibly can but in a case like this, it took human eyes to get a handle on what was going on. However, the post did lead to some positive things that we can test to try and better the classification.

I hope this helps some.

[mikeshand]

Thankyou. That makes a lot of sense. And good detective work.

Mike

[mikeshand]

Unfortunately it still seems to be reporting that second hop failure as “beyond”

2020-11-11 23:09:05: Outage occurred lasting 1 minute 11 seconds , indicating a problem beyond your internet provider.

2020-11-11 22:58:55: Outage occurred lasting 1 minute 9 seconds , indicating a problem beyond your internet provider.

2020-11-11 22:16:18: Outage occurred lasting 1 minute 14 seconds , indicating a problem beyond your internet provider.

And these are for hop 2 64.24.254.203 and 204 this time

Did I have to reset something to make the new algorithm work?

[OutagesIO_Support]

I'll check. Maybe the fix was not sent to production yesterday if they were working on other things.