Using One-Sided RDMA Reads to Build a Fast, CPU-Efficient Key-Value Store

By Christopher Mitchell, Yifeng Geng, Jinyang Li. USENIX Annual Technical Conference, 2013

456 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/e3dfab0c89065c8f66840d1c5b8870eed4cf10ed

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

RDMA over Commodity Ethernet at Scale (2016, methodology): “As we have shown in RDMA has been used to build systems including storage, key-value stores, and distributed transaction systems [17, 26, 28, 37].”
Congestion Control for Large-Scale RDMA Deployments (2015, background): “Other applications such as distributed memory caches [10, 30] and large-scale machine learning demand ultra-low latency (less than 10 μs per hop) message transfers.”
A Unified Architecture for Accelerating Distributed DNN Training in Heterogeneous GPU/CPU Clusters (2020, background): “Conventional one-sided RDMA operations (WRITE and READ) require at least two roundtrips: getting the remote address, and writing (reading) the value to (from) that address [39, 40, 50, 70].”
In-Memory Big Data Management and Processing: A Survey (2015, background): “[252], document stores such as MongoDB [65], Couchbase”
Design Guidelines for High Performance RDMA Systems (2016, methodology): “For example, clients can access a key-value store either by READing directly from the server’s memory [24, 13, 30, 28], or via RPCs as in HERD [20].”
Efficient Memory Disaggregation with Infiniswap (2017, methodology): “With the advent of RDMA, there has been a renewed interest in DSM research, especially via the key-value interface [35, 51, 60, 63, 66, 69].”
Using RDMA efficiently for key-value services (2014, methodology): “RDMA is increasing its presence in datacenters as the hardware becomes cheaper [21].”
Fast in-memory transaction processing using RDMA and HTM (2015, methodology): “To bridge HTM (which essentially uses OCC) and 2PL, DrTM implements the exclusive and shared locks using one-sided RDMA operations, which are cache-coherent with local accesses and thus provide strong consistency with HTM.”
No compromises: distributed transactions with consistency, availability, and performance (2015, unknown): “[13] DECANDIA, G., HASTORUN, D., JAMPANI, M., KAKULAPATI, G., LAKSHMAN, A., PILCHIN, A., SIVASUBRAMANIAN, S., VOSSHALL, P., AND VOGELS, W. Dynamo: Amazon’s highly available key-value store.”
Datacenter RPCs can be General and Fast (2018, background): “Co-design with RDMA is popular, with numerous examples from key-value stores [44, 36, 24, 61, 60], state machine replication [52], and transaction processing systems [25, 61, 20].”
IX: A Protected Dataplane Operating System for High Throughput and Low Latency (2014, methodology): “USENIX Association 11th USENIX Symposium on Operating Systems Design and Implementation (OSDI ’14) 49”