The size and amount of e-science data sets is growing rapidly. Keeping up with the network demand in order to transfer these data sets over the Internet is a challenge. Single links do not have enough capacity anymore. Therefore we need to install more interfaces in the servers and use all available paths in the network. In this paper we describe two new technologies that help to optimally use the capacity of all multiple paths simultaneously. OpenFlow is used to discover the topology of the network,calculate multiple paths and configure those paths on the OpenFlow network. Multipath TCP (MPTCP) is used on the servers to distribute the load across the paths. We describe the end-to-end good put measurements we did in our OpenFlow test bed. We show that we reach a much higher throughput with multiple paths compared to a single path.