What would be a good design for federation servers in our deployment?
We have a production environment and a staging/QE environment. We'd like the production environment to have a federation server on the public net and the staging/QE environment to have a federation server on the testnet. Is this possible?
The only job of our federation servers is the translation of email-like addresses into account ids. We'd like
user*product.com to live on the public net only and
user*staging.product.com to live on the testnet only. Is that possible?
If not, what would be a good design to work around this? The federation server doesn't know if the request came from the public or the testnet. Should we simply return testnet account ids on staging and public account ids on production and not worry about any side effects (e.g. people on testnet sending to public account ids and vice versa)?