On 15 Apr 2011, at 08:45, Adrian Kennard wrote: > Neil J. McRae wrote: >> On 15 Apr 2011, at 07:36, Adrian Kennard <[email protected]> wrote: >>> On 15/04/11 06:47, Neil J. McRae wrote: >>>> On 14 Apr 2011, at 19:08, Adrian Kennard <<mailto:[email protected]>[email protected] >>>> <mailto:[email protected]>> wrote: >>>> >>>>> It is all sorts - we have been dealing with many ISPs of late and a >>>>> scary number do not understand what goes on under to hood! >>>> so is this a general clue problem? In which case this might be useful to >>>> play to a wider audience. >>> I think so - we have had fun explaining RADIUS configs to people quite a >>> few times >>> >>> Three seemed confused over what to expect on an initial RADIUS request, >>> and then confused about what they would see in PPP negotiation later for >>> routed IPs. That sort of thing. >> I find that incredible as its not that complicated. > > I know that! What has actually surprised us (and I am not going to name > names) is that there seem to be ISPs out there that just buy boxes and plug > them together and cross their fingers. To me it seems odd not to understand > the workings in detail, and some say we go a tad far making our own routers, > but at least understanding basics like this is generally quite important.
Yep it is amazing the number of people who cant manage to send us only the attributes we tell them too. We have now taken to filtering out everything we don't want to see. There really is no point in sending me a Framed-IP-Address and a Framed-Route on a session that you want L2tp forwarded especially when you also forget to set Tunnel-Server-Endpoint etc. Neil it would be good if IPSC allowed us to specify the Tunnel-Server-Endpoint via radius, instead of the evil fixed set of IP addresses we currently have to specify on a CRF, so much more flexible. WBC already works like this. But it seams that BT have 2 separate teams looking after radius for IPSC and WBC, I had several meetings with the IPSC design guys before the product was launched (we were trialists) and they could not understand what we were asking for. Steve -- Steve Lalonde RTFM. Chief Technical Officer Entanet International Ltd http://www.enta.net/
