Push-Pull Messaging: a high-performance communication mechanism for commodity SMP clusters

Abstract

Push-Pull Messaging is a novel messaging mechanism for high-speed interprocess communication in a cluster of symmetric multi-processors (SMP) machines. This messaging mechanism exploits the parallelism in SMP nodes by allowing the execution of communication stages of a messaging event on different processors to achieve maximum performance. Push-Pull Messaging facilitates further improvement on communication performance by employing three optimizing techniques in our design: (1) Cross-Space Zero Buffer provides a unified buffer management mechanism to achieve a copy-less communication for the data transfer among processes within a SMP node. (2) Address Translation Overhead Masking removes the address translation overhead from the critical path in the internode communication. (3) Push-and-Acknowledge Overlapping overlaps the push and acknowledge phases to hide the acknowledge latency. Overall, Push-Pull Messaging effectively utilizes the system resources and improves the communication speed. It has been implemented to support high-speed communication for connecting quad Pentium Pro SMPs with 100 Mbit/s Fast Ethernet.