Cutting tail latency in cloud data stores via adaptive replica selection

February 6, 2015
Marco Canini | Université catholique de Louvain

Achieving predictable performance is critical for many distributed applications, yet difficult to achieve due to many factors that skew the tail of the latency distribution even in well-provisioned systems. In this talk, we will present the fundamental challenges involved in designing a replica selection scheme that is robust in the face of performance fluctuations across servers. We will then present the design and implementation of an adaptive replica selection mechanism, C3, that is robust to performance variability in the environment. We will also discuss C3’s effectiveness in reducing the latency tail and improving throughput through results of performance evaluations conducted on Amazon EC2 and through simulations. An implementation of C3 inside Cassandra improved tail latencies by factors exceeding 3x, while also improving throughput by up to 50%.

Speaker Details

Marco is an assistant professor in the ICTEAM institute at the Université catholique de Louvain. Marco obtained his Ph.D. in computer science and engineering from the University of Genoa in 2009 after spending the last year as a visiting student at the University of Cambridge, Computer Laboratory. He holds a laurea degree with honors in computer science and engineering from the University of Genoa. He was a postdoctoral researcher at EPFL from 2009 to 2012 and after that a senior research scientist for one year at Deutsche Telekom Innovation Labs & TU Berlin. He also held positions at Intel Research and Google.