I was actually referring to Plus.net having a backup service of Telehouse North somewhere else so that the telphones plus other services didn't fail like they did.

I asked the same over on PUG and apparently they do but for some reason BT's config was only using the Radius in THN. The delay in using the THE Radius was caused by getting BT to amend their config. In future it should help ease Radius load as well because BT should now be using more servers for the Radius authentication.

In reply to:

---

You would hope pretty much everything in Telehouse has a redundant hot-swap on standby, seeing as its so important and all...

---

I met the building manager for Telehouse yesterday as it happens and had a long discussion about various issues (unrelated to PN I should add).. Telehouse are clued up. (We host servers in Telehouse and have never had a problem, but this doesn't mean that we've done anything better as the problems are often localised in Telehouse--which is a good thing.. Other datacentres lose power to entire floors or half the building... or even dare I say the entire building.. or have water leaks etc..

Getting redundant dual feeds in some parts of Telehouse (such as the one we're in) is not possible.. Some datacentres will give you two feeds but they aren't redundant as you will notice when things go wrong.. Also they need to weigh the costs in proportion to the benefits.. Having small problems is not a major issue in comparison to hours and hours of downtime..

This isn't to say PlusNet can't buy UPSs for their routers and DSL termination kit (we have them on our routers.. but not AG webservers for example... we rely on Telehouse's UPSs for that).. but everything needs to be in proportion to the effects it will have.. DSL customers being offline for a few hours a year (which is probably an over estimation of average effects of problems) is not a major issue.. since DSL isn't really (at least in the way PN package it) meant to be a critical service. (reference to PN could just as well be to most ISPs.. no inference on PN..)

seb

"But they have, they have centrals terminating in THE"

3 out of the 5 centrals plus the Tiscali LLU connection are in Telehouse North.

"What I fail to understand is why they have all their radius servers in THN"

They don't. IIRC, they have a cluster in each building plus a backup cluster they can utilise in Sheffield. It is possible one cluster got swamped, in which case they need to increase the processing capacity of each cluster.

TT

In reply to:

---

It is possible one cluster got swamped, in which case they need to increase the processing capacity of each cluster.

---

While I appreciate PN have around 10 times the number of customers we deal with, we look after the Radius servers for a mobile network for one of the value added products they use on it. We have around 20k customers on radius servers (two in two locations). Radius handles auth and accounting, as well as IP block assignment, much as any ISP would. And to be quite frank, the servers are P3 / 800's and I don't believe I've ever seen then run above about 5% unless we are hacking the hell out of them running reporting. While PN may well have 200k customers, theres no genius required to see that any modern commercial server would handle the load adequately...

And I also appreciate it may be an issue of 100k connections landing at once, but we've had GGSM outages with the mobile network that have caused 10k clients to appear at once. They were all authed within a couple of mins, even on these relatively shoddy servers.

Me wonders exactly what junk PN are using for Radius "Clusters"...

In reply to:

---

They don't. IIRC, they have a cluster in each building plus a backup cluster they can utilise in Sheffield. It is possible one cluster got swamped, in which case they need to increase the processing capacity of each cluster.

---

_TT_, according to Simon on PUG, they weren't getting swamped because BT weren't sending any radius requests to them due to a config error on their part. Once BT had corrected this problem then radius requests were processed by the THE radius and the remaining 'up' centrals filled up fairly quickly.

I distinctly remember telling Plusnet originally that the phone system they were setting up had a serious flaw in it's redundancy. And guess what, thats the second or third time now I've been proven correct.

Although it is extremely cunning - in an outage, they can't get overwhelmed with support because you just can't get through. It certainly solves that problem!

In reply to:

---

our RADIUS servers

---

REDUNDANCY IS WHERE? You'd think after the last radius problems Plusnet would have re-engineered the RADIUS service completely. But instead, they left it all in one physical site. Genius. Absolute genius.

In reply to:

---

At approximately 11am we were able to revert to our backup telephone system, which is allowing our customers to call the Customer Support Centre, although the Interactive Voice Response (IVR) will not be operational.

---

And the automatic failover is...?

In reply to:

---

TT_, according to Simon on PUG, they weren't getting swamped because BT weren't sending any radius requests to them due to a config error on their part. Once BT had corrected this problem then radius requests were processed by the THE radius and the remaining 'up' centrals filled up fairly quickly.

---

So either Plusnet, or BT, failed to test this setup then I guess...?

Well to be fair nobody has any need to post in it - it's not an official (or recognised PlusNet support forum) and although the negative postings may not help PlusNet users they do serve to highlight the issues for any potential customers they may visit the forum to 'test the water' before signing up.

The truth is that some people get too much amusement from poking fun at other ISP's and other people get too upset and defensive of their chosen ISP, probably out of frustration. As for lugz and others making the Plusnet staff leave - if you believe that you need your head examined.

If you have a look through the forum most of the posts are by customers with problems and poor/non existant support. Some folks just like to target the minority of naysayers and blame them - when the truth is if they weren't here the forum would still be full of PlusNet customers having problems and poor/non-existant support as that's the way Plusnet are these days.

The reason people are here poking fun is because PlusNet are an easy target. They are an easy target as they have so many serious problems of their own making.....

Anyway - why does Dusty blame the Elk - what have they done?