How problematic are the recent LNS issues in practice?

Last week I was about to pull the trigger on becoming an AAISP customer when the stability issues caused by their core equipment gave me pause. I'm happy to pay a premium for good service, but I define it as "doesn't fail that often" instead of "highly competent people get elbow deep into hardware when it fails repeatedly".

So, what do these issues look like from a customer perspective? If an LNS falls over, is there reasonably quick and seamless failover to another piece of hardware, or does it take a few minutes for failure detection and reauthentication, or does it take manual intervention and longer than a few minutes?

The first I could live with, the second and third alternative IMHO aren't premium service.

Experiences / opinions would be appreciated.

Edited by qazwsxedc (Sun 28-Jan-24 10:43:26)

The recent issues have only affected those on speeds >80Meg, so if you are on VDSL you wouldn't have been affected.

As soon as one of these hardware lockups happen it is possible to immediately reconnect, with the connection going to an alternative LNS.

How it looks to you as a customer will vary depending on how your router responds to a sudden drop and they all tend to vary. Best case scenario you might not even notice the connection had dropped if your router immediately attempts to reconnect, worst case scenario you need to manually intervene and reboot the router (unlikely). Some devices might seem offline a bit longer depending on how the router deals with firewall states.

Even with a quick reconnect it probably isn't quick enough to avoid a glitch or drop of ongoing video/audio call or streaming service, but this isn't something that is happening every day, plus it will get fixed.

Edited by E300 (Mon 29-Jan-24 08:44:09)

Hi,

One loss for me ballsed up a large overnight download.
Another dropped a Teams call for the Mrs
A more recent one resulted in downtime as PPP didn't reconnect automatically.

---

Also - I notice slightly lower latency on y.witless than the new all singing z.witless... which is a step backwards

Edited by perlen (Mon 29-Jan-24 18:22:36)

Unless you do a big load of connect / disconnect cycles to the two different LNS to establish an average baseline latency, your variation might be in the BTW network. I've always seen a variation of up to 2ms on my baseline latency just by disconnecting the PPP session and re-connecting a bunch of times, both with AAISP (previous ISP) and UnchainedISP (current). As far as I can see, there seem to be faster paths through the BTW network, and whether you end up on them is pot luck. There's no-way to know exactly what is going on as we can't see any details from BTW as to how the PPP session actually gets to the destination - it's all a black box at that point.

I'm on VDSL so haven't been affected by the recent problems but normally when an LNS falls over you will get an immediate disconnect. It is then normally ~1-2 minute or so for a reconnect depending on equipment your end.

Never had a scenario myself where I've had to manually intervene.

I would give them a go, it is likely they will get to the bottom of this issue soon-ish and you get the added benefit of some transparency.

Most of the larger ISPs likely have reconnects too (admittedly probably not quite as many as AAISP) but don't publish or due to the larger customer base it is not pinpointed as accurately by customers.

I've always found Z.Witless adds around 1.5 to 2ms on my minimum latency, I believe it is physically in a different location (different data centre) than x and y so just gets a different routing on BT Wholesale I suppose. Its possible to drop and reconnect a few times and get a better routing but it is always higher than I can get on X or Y. Just now I dropped and reconnected to X, was 8.7ms latency to the LNS and on reconnecting I got lucky and its dropped to 6.2ms. I've had similar experiences on all ISPs using BT Wholesale.

I dropped the connection just now to see if I would come back up on the new G.Gormless LNS which is now installed, but I came back up on the same one but did get the better routing.

It's a shame we can't steer ourselves to an LNS of our own choosing, I know we can get to the test LNS by prefixing with "test-", but as far as I'm aware we can't do that with the production LNS's, perhaps that might cause issues where we all end favouring one over the others.

In reply to a post by E300:

It's a shame we can't steer ourselves to an LNS of our own choosing

Mikrotik lets you select the access concentrator name in their PPPoE client settings.

I've never tried forcing selection of one particular access concentrator, but in principle if you get PADO offers from more than one, it should only honour those from the selected one.

Downside would be, if that particular LNS goes down, you won't be able to establish a connection.