2021/09/20 by Hammurabi Mendes, Mendes, Hammurabi, Bryce Wiedenbeck +3
Computer Science · #Asynchronous communication #Cloud Computing and Resource Management #Cloud computing #Computer network #Computer science #Distributed #Distributed and Parallel Computing Systems #Distributed computing #FOS: Computer and information sciences #Middleware (distributed applications) #Networking and Internet Architecture (cs.NI) #Operating system #Overhead (engineering) #Parallel #Peer-to-Peer Network Technologies #Remote direct memory access #Remote procedure call #Tree (set theory) #Usability #and Cluster Computing (cs.DC)
paper · pdf · doi:10.48550/arxiv.2109.09819
openalex publication_date 2021/09/20 · openalex created_date 2022/07/25 · openalex updated_date 2026/07/28
We introduce Seriema, a middleware that integrates RDMA-based remote\ninvocation, asynchronous data transfer, NUMA-aware automatic management of\nregistered memory, and message aggregation in idiomatic C++1x. Seriema supports\nthe notion that remote invocation and asynchronous data transfer are\ncomplementary services that, when tightly-integrated, allow distributed data\nstructures, overlay networks, and Cloud & datacenter service applications to be\nexpressed effectively and naturally, resembling sequential code. In order to\nevaluate the usability of Seriema, we implement a Monte-Carlo Tree Search\n(MCTS) application framework, which runs distributed simulations given only a\nsequential problem specification. Micro-benchmarks show that Seriema provides\nremote invocations with low overhead, and that our MCTS application framework\nscales well up to the number of non-hyperthreaded CPU cores while simulating\nplays of the board game Hex.\n