On Accelerating Distributed Convex Optimizations