In the distributed FSF we want to provide the same level of support for spare capacity sharing that is provided for processing nodes. This is a difficult task in the case of a distributed system, because the decisions made in one node may affect another one, requiring distributed consensus. For example, a distributed transaction may have several activities executing in different processing nodes. One of them is periodic, and the others are activated by the arrival of a message from the preceding activity. Therefore, the latter activities inherit the period of the first activity (with the additional jitter introduced by the processing and message transmission). If the transaction allows a continuous scale of periods between some minimum and maximum values, separate negotiations in the network and in the different processing nodes will most probably result in different periods because the spare capacity is different in each node. Since the transaction cannot run with different periods, there needs to be some renegotiation to change the period to the maximum obtained (representing the minimum resource consumption). During this renegotiation things might have changed, requiring further renegotiation rounds.