[OpenSIPS-Users] opensips-cp CDR correlation

[Iñaki Baz Castillo]

2009/4/30 Bogdan-Andrei Iancu <bogdan at voice-system.ro>:
>> Yes, I agree on it. However I just wanted to mean that using a
>> media-proxy is not the best solution for all the cases, specially when
>> clients are behind same NAT (an office for example)
>
> from the open internet, you cannot tell (100% sure) if two devices are in
> the same subnet! it is a bit risky.

Sure, of course this should be configurable by the provider since it
could know the client topology.
Relying on received public address is not a good idea in those
countries using double-NAT XDD



> I tend to agree SST is more efficient, but :

> (1) less accurate (20 minute
> timer has a good error :D) ,

I'm not an expert in SST, but AFAIR interfavl checking is
configurable, so 30-60 seconds would be enough accurate.


> (2) not widely supported by all devices and

SST just require *one* endpoint implementing them in order to work
(the endpoint forcing SST sends a re-INVITE/UPDATE to the other
endpoint). For this, we need the PSTN gateways supporting SST or using
a pure SIP B2BUA which performs SST in both legs.


> (3) exposed to hacks (as it is signalling, the same as acc).

Not if the accouting is done in a pure SIP B2BUA :)


Regards.

-- 
Iñaki Baz Castillo
<ibc at aliax.net>