Abstract
QuA is a reflective middleware architecture with a implementation broker assisting the service planner in service planning by performing resource discovery. QuA supports pluggable core services and one of these services is the peer-to-peer broker. The peer-to-peer broker is a distributed approach to the normal server/client way of managing resource discovery. The resources are distributed among the participants in the peer-to-peer network making the network more resilient to resource loss than the normal client/server approach.
This thesis evaluates the peer-to-peer broker by looking at the distribution of resources and the disk space and bandwidth used by this. By making the searchable domain for each resource larger by using more disk space we see if the resource discovery time can be improved and how big the disk and bandwidth overhead is by doing this.
The evaluation of the peer-to-peer broker shows that by partitioning the searchable domain we do get some improvements in the resource discovery time, but at the expense of a large disk space usage and bandwidth overhead. Also the size of the searchable domain for a resource can become very large.