Using RDMA Efficiently for Key-Value Services

被引:15
作者
Kalia, Anuj [1 ]
Kaminsky, Michael [2 ]
Andersen, David G. [1 ]
机构
[1] Carnegie Mellon Univ, Pittsburgh, PA 15213 USA
[2] Intel Labs, Santa Clara, CA USA
关键词
RDMA; InfiniBand; RoCE; Key-Value Stores;
D O I
10.1145/2740070.2626299
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
This paper describes the design and implementation of HERD, a key-value system designed to make the best use of an RDMA network. Unlike prior RDMA-based key-value systems, HERD focuses its design on reducing network round trips while using efficient RDMA primitives; the result is substantially lower latency, and throughput that saturates modern, commodity RDMA hardware. HERD has two unconventional decisions: First, it does not use RDMA reads, despite the allure of operations that bypass the remote CPU entirely. Second, it uses a mix of RDMA and messaging verbs, despite the conventional wisdom that the messaging primitives are slow. A HERD client writes its request into the server's memory; the server computes the reply. This design uses a single round trip for all requests and supports up to 26 million key-value operations per second with 5 mu s average latency. Notably, for small key-value items, our full system throughput is similar to native RDMA read throughput and is over 2X higher than recent RDMA-based key-value systems. We believe that HERD further serves as an effective template for the construction of RDMA-based datacenter services.
引用
收藏
页码:295 / 306
页数:12
相关论文
共 28 条
[1]  
[Anonymous], 2014, PROC USENIX NSDI
[2]  
Atikoglu Berk, 2012, Performance Evaluation Review, V40, P53, DOI 10.1145/2318857.2254766
[3]  
Cooper B. F., 2010, P ACM S CLOUD COMP S, P143, DOI DOI 10.1145/1807128.1807152
[4]  
Dragojevic Aleksandar, 2014, USENIX NSDI
[5]  
Fan D. G., 2013, P 10 USENIX C NETW S, P371
[6]  
Flajslik M., 2013, USENIX ANN TECHNICAL
[7]  
Gibson G., PROBE 1000 NODE EXPT
[8]  
Herlihy M., 2008, DISC
[9]  
Huang J., 2012, IPDPS
[10]  
Jose J., 2012, CCGRID IEEE