Fast in-memory transaction processing using RDMA and HTM

By Xingda Wei, Jiaxin Shi, Yanzhe Chen, Rong Chen, Haibo Chen. Symposium on Operating Systems Principles, 2015

341 citations recorded by the source corpus.

How researchers see this paper

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

Design Guidelines for High Performance RDMA Systems (2016, background): “Finally, we show that today's RDMA NICs handle contention for atomic operations extremely slowly, rendering designs that use them [27] [30] [11] very slow.”
NetChain: Scale-Free Sub-RTT Coordination (2018, background): “State-of-the-art in-memory transaction processing systems such as FaRM [5] and DrTM [6], which can process hundreds of millions of transactions per second with a latency of tens of microseconds, crucially depend on fast distributed locking to mediate concurrent access to data partitioned in multiple”
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].”
Datacenter RPCs can be General and Fast (2018, methodology): “The result has been an explosion of co-designed distributed systems that depend on niche network technologies, including RDMA [18, 25, 26, 38, 50, 51, 58, 64, 66, 69], lossless networks [39, 47], FPGAs [33, 34], and programmable switches [37].”