Not So Synchronous RPC: RPC with Silent Synchrony Switch for Avoiding Repeated Marshalling of Data

Abstract

The Remote Procedure Call (RPC) is a powerful technique for constructing distributed, client-server based applications. In traditional RPC systems, when a remote procedure call fails to return result to the client because of communication failure, client stub usually reinvokes the remote method for expected output if the method is idempotent. This results in repeated marshalling of same data multiple times which is a burden over network if the data is huge. For the first time, we prevent this repeated marshalling by introducing a behavioral pattern that we refer to as usually synchronous, but conditionally asynchronous RPC (CA-RPC). CA-RPC establishes handshaking between client stub and server stub to understand whether connection was failed after demarshalling of data at server stub. If so, the client stub runs a pulling thread to retrieve the result without remarshalling the same data. Besides that, our strategy enhances the performance by a synchrony switch at runtime under some condition. Our results show that CA-RPC has better performance than traditional RPC for cases where data marshalling has significant overhead over low rate data connections.