Parallel server systems model customer call centers and web server farms. We focus on these systems when the offered load is high, and consequently the number of servers is large. In this talk, we will first overview this active research area. Next, we will consider a parallel server system that consists of several customer classes and server pools in parallel. We propose a simple robust control policy to minimize the total linear holding and reneging costs. We show that this policy is asymptotically optimal under the many-server heavy traffic for parallel server systems when the service times are only server pool dependent and exponentially distributed. The talk is based on joint work with Tolga Tezcan of University of Illinois at Urbana-Champaign.