The scalability of FSF has to be analyzed from two points of view. From the execution runtime overhead, the scheduling algorithms used to manage the server information and schedule the threads have O (log n) complexity, where n is the number of servers currently instanced in the system. Adding the server management scheme on top of the EDF or fixed priority scheduler does not add further complexity to the algorithm, so the framework is very scalable. From the point of view of negotiation times, the scalability depends on the particular admission test implemented. For example, in Shark we implemented an utilization based test that had the restriction of deadlines equal to periods, and the advantage of a linear dependency, O (n), on the number of servers. In MaRTE OS we used a more elaborate utilization test that can be used for deadlines less than or equal to the periods, which has a time dependency of O (n2), making it a bit less scalable. Of course, off-line admission test may be carried out for a fully scalable approach.