Remote direct memory access socket aggregation

摘要

Byte utilization is improved in Remote Direct Memory Access (RDMA) communications by detecting a plurality of concurrent messages on a plurality of application sockets which are destined for the same application, client or computer, intercepting those messages and consolidating their payloads into larger payloads, and then transmitting those consolidated messages to the destination, thereby increasing the payload-to-overhead byte utilization of the RDMA transmissions. At the receiving end, multiplexing information is used to unpack the consolidated messages, and to put the original payloads into a plurality of messages which are then fed into the receiving sockets to the destination application, client or computer, thereby making the consolidation process transparent between the initiator and the target.

主权项

1. A method for providing remote direct memory access communication comprising:
at a first data processing system, intercepting a plurality of concurrent application socket messages prior to transmission to a common application socket destination, each intercepted message having a payload; still prior to transmission to the common destination, aggregating by the first data processing system the payloads of the intercepted concurrent application socket messages into one or more consolidated payloads of one or more application socket messages; notifying by the first data processing system the common application socket destination of the aggregation and multiplexing information using an out-of-band communication; and transmitting by the first data processing system to the common application socket destination the one or more consolidated payloads in the one or more application socket messages on fewer sockets than the plurality of intercepted application socket messages, wherein a ratio of payload to protocol overhead byte utilization is increased compared to the same ratio for the plurality of intercepted messages; wherein the intercepting and aggregating are performed in a communications stack of the first data processing system at a level compliant with Open Systems Interface level 5 session level, and wherein the first data processing system comprises a computer having a processor.