Major Description
To work efficiently, we should have one (or several) accept threads per NUMA node. In addition, it should schedule the VCs on ET_NET threads that are assigned to the same NUMA node. This assures that memory allocated for the VC stays on the same NUMA node.