The last third of the lectures on Parallel Computing deals with characteristics of distributed memory systems where processors or shared memory processor-nodes are interconnected via a communication network which has to be used to exchange information between processes or threads in any non-trivial parallel algorithm. The concept of message-passing as a means to structure and implement algorithms is introduced, concretely via MPI which is dealt with extensively, regarding both fundamental message-passing ideas but also the way these have been realized in a concrete specification.