Remote Batch Invocation for Compositional Object Services

Abstract

Abstract. Because Remote Procedure Calls do not compose efficiently, designers of distributed object systems use Data Transfer and Remote Façade patterns to create large-granularity interfaces, hard-coded for particular client use cases. As an alternative to RPC-based distributed objects, this paper presents Remote Batch Invocation (RBI), language support for explicit client-defined batches. A Remote Batch statement combines remote and local execution: all the remote code is executed in a single round-trip to the server, where all data sent to the server and results from the batch are communicated in bulk. RBI supports remote blocks, iteration and conditionals, and local handling of remote exceptions. RBI is efficient even for fine-grained interfaces, eliminating the need for hand-optimized server interfaces. We demonstrate RBI with an extension to Java, using RMI internally as the transport layer. RBI supports large-granularity, stateless server interactions, characteristic of service-oriented computing.

Efficient Session Type Guided Distributed Interaction

“…Hence, we implement a flow sensitive data flow analysis in the compiler which determines whether the loop conditional or the sends are dependent on any of the receives in the loop body. This analysis is similar to the one described in remote batch invocation [12]. Session type information allows us to precisely determine the scope of the analysis.…”

“…Any operation performed on the result of remote method calls forces the method call to be flushed. Remote batch invocation (RBI) [12] performs batching on a block of code marked by the programmer. RBI reorders operations such that all remote operations are performed after the local operations and the code is exported to the server.…”

“…To make matters worse, specialization is not even possible if an RMI call depends on local operations. Previous work has looked at alleviating some of these costs through various techniques, but are limited to bi-party interaction, semantically restricted to part of the protocol [12], or require code to be written in a style amenable to enhancement [13].…”

“…We are very grateful to A. Ibrahim, Y. Jiao, E. Tilevich, and W. Cook for sharing insights and the source code for RBI [12] with us.…”

Effective fusion and separation of distribution, fault-tolerance, and energy-efficiency concerns

“…We have implemented three Lucene-based services using five different middleware platforms: TCP sockets, synchronous and asynchronous remote calls in R-OSGi [120], Message Oriented Middleware (MOM) [9], and Remote Batch Invocation (RBI) [61]. For each implementation, we measured: (1) the total number of lines of uncommented code and its cyclomatic complexity, (2) the aggregate latency of invoking remote service methods, and (3) the degree of reliability of remote service methods in the presence of network volatility.…”