We build web services that are for consumption by known third-parties.
We tend to implement very basic security that involves:
- A known token ID which we communicate to both parties
- Restrict access to known subset of IP addresses
- Secure the transport layer via SSL
I’m not comfortable with this, but implementing Federated Security (using WIF/ADFS 2) is VERY complex! How can I justify using this tech – what is fundamentally wrong with the above approach?
I realise that the web service (which might expose sensitive data) is now only as secure as the token, but so would a username/password combo.
Thanks
Duncan
To improve or replace the common token i would use some client/server certificate authentification. You will use SSL anyway and client certificate gives some more features