Mitigating Load Imbalance in Distributed Data Serving with Rack-Scale Memory Pooling
ACM Transactions on Computer Systems · 31.05.2018 · 36 (2) · 1-37
DOI: 10.1145/3309986 ↗Kurz gefasst
Fachartikel in der Zeitschrift ACM Transactions on Computer Systems, online erschienen am 09.04.2019. Verfasst von Stanko Novakovic und 5 weiteren Autor:innen. Die Arbeit wurde bisher 10-mal zitiert (Crossref) und nennt 73 Literaturangaben. Gefördert von: Nano-Tera YINS project, CHIST-ERA DIVIDEND project, Scale-Out NUMA project of the Microsoft-EPFL Joint Research Center.
Abstract (Originalsprache)
To provide low-latency and high-throughput guarantees, most large key-value stores keep the data in the memory of many servers. Despite the natural parallelism across lookups, the load imbalance, introduced by heavy skew in the popularity distribution of keys, limits performance. To avoid violating tail latency service-level objectives, systems tend to keep server utilization low and organize the data in micro-shards, which provides units of migration and replication for the purpose of load balancing. These techniques reduce the skew but incur additional monitoring, data replication, and consistency maintenance overheads. In this work, we introduce RackOut, a memory pooling technique that leverages the one-sided remote read primitive of emerging rack-scale systems to mitigate load imbalance while respecting service-level objectives. In RackOut, the data are aggregated at rack-scale granularity, with all of the participating servers in the rack jointly servicing all of the rack’s micro-shards. We develop a queuing model to evaluate the impact of RackOut at the datacenter scale. In addition, we implement a RackOut proof-of-concept key-value store, evaluate it on two experimental platforms based on RDMA and Scale-Out NUMA, and use these results to validate the model. We devise two distinct approaches to load balancing within a RackOut unit, one based on random selection of nodes—RackOut_static—and another one based on an adaptive load balancing mechanism—RackOut_adaptive. Our results show that RackOut_static increases throughput by up to 6× for RDMA and 8.6× for Scale-Out NUMA compared to a scale-out deployment, while respecting tight tail latency service-level objectives. RackOut_adaptive improves the throughput by 30% for workloads with 20% of writes over RackOut_static.
Förderung
- Nano-Tera YINS project
- CHIST-ERA DIVIDEND project
- Scale-Out NUMA project of the Microsoft-EPFL Joint Research Center
Zitieren
APA
Novakovic, S., Daglis, A., Ustiugov, D., Bugnion, E., Falsafi, B., & Grot, B. (2018). Mitigating Load Imbalance in Distributed Data Serving with Rack-Scale Memory Pooling. ACM Transactions on Computer Systems, 36(2), 1-37. https://doi.org/10.1145/3309986
Vancouver
Novakovic S, Daglis A, Ustiugov D, Bugnion E, Falsafi B, Grot B. Mitigating Load Imbalance in Distributed Data Serving with Rack-Scale Memory Pooling. ACM Transactions on Computer Systems. 2018;36(2):1-37. doi:10.1145/3309986
BibTeX
@article{Novakovic2018,
author = {Novakovic, Stanko and Daglis, Alexandros and Ustiugov, Dmitrii and Bugnion, Edouard and Falsafi, Babak and Grot, Boris},
title = {Mitigating Load Imbalance in Distributed Data Serving with Rack-Scale Memory Pooling},
journal = {ACM Transactions on Computer Systems},
year = {2018},
volume = {36},
number = {2},
pages = {1--37},
doi = {10.1145/3309986}
}
EUVIMED