Seriema: RDMA-based Remote Invocationwith a Case-Study on Monte-Carlo Tree Search

TL;DR

Seriema is a middleware that combines RDMA-based remote invocation with asynchronous data transfer and memory management, enabling efficient distributed data structures and applications, demonstrated through a scalable Monte-Carlo Tree Search case study.

Contribution

Introducing Seriema, a middleware integrating RDMA remote invocation, asynchronous transfer, and memory management, facilitating natural expression of distributed applications in C++.

Findings

Low-overhead remote invocation performance

Scales well for distributed Monte-Carlo Tree Search

Effective implementation of distributed simulations for Hex

Abstract

We introduce Seriema, a middleware that integrates RDMA-based remote invocation, asynchronous data transfer, NUMA-aware automatic management of registered memory, and message aggregation in idiomatic C++1x. Seriema supports the notion that remote invocation and asynchronous data transfer are complementary services that, when tightly-integrated, allow distributed data structures, overlay networks, and Cloud & datacenter service applications to be expressed effectively and naturally, resembling sequential code. In order to evaluate the usability of Seriema, we implement a Monte-Carlo Tree Search (MCTS) application framework, which runs distributed simulations given only a sequential problem specification. Micro-benchmarks show that Seriema provides remote invocations with low overhead, and that our MCTS application framework scales well up to the number of non-hyperthreaded CPU cores while simulating plays of the board game Hex.