If an application uses multiple farms simultaneously, then a single farm daemon could become a bottleneck. In that case, a farm daemon for every farm might give better results. This would need to be verified since the additional tasks could have an overhead in the sense that there would be more tasks on a fixed number of nodes in the network. Also, since farms are based on a named service, an additional name server task would need to be introduced (if two tasks were to simultaneously request for creating a farm with the same name, one would get an error and the other would be allowed to continue). It would need to be studied whether adding extra farm daemons would improve the performance and how often would a large number of farms would be used in an application. This could be used for future work in this area.