Is it possible to use FreeTDS to implement a stub TDS server?

Let's say that I have a black-box application which produces a
massive, continuous stream of data and dumps it into an MS SQL Server
database.  I want to pull the data *out* of this database as quickly
as it is inserted, and pass it on for further processing in
near-real-time.

In fact, I have no interest in keeping the data in a relational
database at all; once it's been extracted I have no further use for
the SQL database, and I don't have disk space to store the data
permanently.  The SQL server is simply acting as a (rather
inefficient) message queue.

So what I would really like would be to avoid the use of MS SQL Server
entirely, and simply connect the application to a "stub" server.  That
server would pretend to be MS SQL Server, and would respond to the
application's requests as a real SQL server would, but in fact would
simply be relaying the inserted data to an AMQP server or similar.

* Has anybody done anything like this before?

* Is it feasible to emulate the behavior of MS SQL Server to the
  degree that a client won't know the difference, assuming that the
  client is treating the server as a write-only output repository?

* Does FreeTDS include low-level protocol code that would be useful in
  implementing such a stub server, or would it be simpler to start
  from scratch?

Benjamin