> internet service provider A offers an e-mail to SMS service. he
> forwards the message from his application to kannel and passes on
> username to his local kannel as "customername". The request would be
> routed to another wholesale provider using kannel as well. There the
> username would appear as "ispname:customername". So if at any time
> there's a query on a message, this mechanism can be easily used to
> pull out the logs. Also the wholesaler can make accounting specific
> to the isp's different customers or services.
If I understand correctly, the message will go
sendsms request -> Kannel -> HTTP SMSC 1 (another Kannel) -> .... -> HTTP
SMSC n (another Kannel) -> SMSC
For the first part ("customername"), do you propose to have a fixed
parameter for that sendsms-user, or a sendsms cgi parameter? Perhaps the
former is better because you'll always get the username, but in that case
you need to configure one sendsms-user per customer. And in the second part,
it's appended onto a parameter ("ispname") defined for the HTTP SMSC?
>
> The thing for this to work is:
>
> a) user accounting only uses username up to the ":" sign to authorize.
> b) original username is passed on appended on the HTTP SMSC (must be
> configurable for security and/or backward compatibility).
Sounds fine, good for a complete audit trail.
