The digital philosophy claims that all physical course of our environment are forms of computation or information processing at the most fundamental level. This concept gives rise to the concept of Big data. As a human nature we all need thing to be fast always. For faster processing we have to rely on parallel processing. This research proposes a new parallel computing architecture called as Extended Double Star (EDS) for implementing the big data framework. The different topological parameters are derived for EDS and compared with the parent networks. Next efficient routing and broadcasting algorithms are proposed to show the efficient message passing using the network controllers. The link complexity of EDS network is found to be least among all extended topologies. While packing the highest number of nodes the EDS network retains all original features of the base network but results in faster routing and broadcasting.