Bubble-flux

2017 IEEE International Symposium on High Performance Computer Architecture (HPCA)

et al. 2017

In 2013, U.S. data centers accounted for 2.2% of the country's total electricity consumption, a figure that is projected to increase rapidly over the next decade. Many important workloads are interactive, and they demand strict levels of quality-of-service (QoS) to meet user expectations, making it challenging to reduce power consumption due to increasing performance demands.This paper introduces Hipster, a technique that combines heuristics and reinforcement learning to manage latency-critical workloads. Hipster's goal is to improve resource efficiency in data centers while respecting the QoS of the latency-critical workloads. Hipster achieves its goal by exploring heterogeneous multicores and dynamic voltage and frequency scaling (DVFS). To improve data center utilization and make best usage of the available resources, Hipster can dynamically assign remaining cores to batch workloads without violating the QoS constraints for the latency-critical workloads. We perform experiments using a 64-bit ARM big.LITTLE platform, and show that, compared to prior work, Hipster improves the QoS guarantee for Web-Search from 80% to 96%, and for Memcached from 92% to 99%, while reducing the energy consumption by up to 18%.

Section: Experimental Methodologymentioning

confidence: 99%

Hipster: Hybrid Task Manager for Latency-Critical Cloud Workloads

2017 IEEE International Symposium on High Performance Computer Architecture (HPCA)

et al. 2017

“…These techniques require extensive microarchitectural feature tuning prior to deployment in production clusters. Secondly, prior work [9,14,16,21,69,70] detect interference among colocated services and adjust resource allocation dynamically or disallow colocation of LC services. While this approach works well to maintain QoS, it is imperative to both increase system throughput and improve energy efficiency to increase revenue.…”

Section: B Twig Evaluationmentioning

confidence: 99%

“…Managing shared-resource contention is a well-studied but still an open problem [5,[8][9][10][11]. Prior work has addressed this problem in two ways, by (a) disallowing resource sharing for LC services in periods of high load to avoid interference [12][13][14][15][16][17][18][19] or (b) disallowing colocation of services even if they are unlikely to interfere with each other [20][21][22]. Blindly applying either solution preserves the QoS of the LC service but results in low server-efficiency.…”

Section: Introductionmentioning

confidence: 99%

Twig: Multi-Agent Task Management for Colocated Latency-Critical Cloud Services

2020 IEEE International Symposium on High Performance Computer Architecture (HPCA)

et al. 2020

Many of the important services running on data centres are latency-critical, time-varying, and demand strict user satisfaction. Stringent tail-latency targets for colocated services and increasing system complexity make it challenging to reduce the power consumption of data centres. Data centres typically sacrifice server efficiency to maintain tail-latency targets resulting in an increased total cost of ownership. This paper introduces Twig, a scalable quality-of-service (QoS) aware task manager for latency-critical services co-located on a server system. Twig successfully leverages deep reinforcement learning to characterise tail latency using hardware performance counters and to drive energy-efficient task management decisions in data centres. We evaluate Twig on a typical data centre server managing four widely used latency-critical services. Our results show that Twig outperforms prior works in reducing energy usage by up to 38% while achieving up to 99% QoS guarantee for latency-critical services.

“…It is con gured to model diurnal load changes (Figure 1), simulating a period of 36 hours [47]; each hour in the original workload corresponds to one minute in our experiments. For the batch workloads [44,71], we use the programs from the SPEC CPU 2006 suite [30].…”

Section: :12 • R Nishtala Et Almentioning

confidence: 99%

The Hipster Approach for Improving Cloud System Efficiency

ACM Trans. Comput. Syst.

et al. 2017

In 2013, U.S. data centers accounted for 2.2% of the country's total electricity consumption, a gure that is projected to increase rapidly over the next decade. Many important data center workloads in cloud computing are interactive, and they demand strict levels of quality-of-service (QoS) to meet user expectations, making it challenging to optimize power consumption along with increasing performance demands. This paper introduces Hipster, a technique that combines heuristics and reinforcement learning to improve resource e ciency in cloud systems. Hipster explores heterogeneous multi-cores and dynamic voltage and frequency scaling (DVFS) for reducing energy consumption while managing the QoS of the latency-critical workloads. To improve data center utilization and make best usage of the available resources, Hipster can dynamically assign remaining cores to batch workloads without violating the QoS constraints for the latency-critical workloads. We perform experiments using a 64-bit ARM big.LITTLE platform, and show that, compared to prior work, Hipster improves the QoS guarantee for Web-Search from 80% to 96%, and for Memcached from 92% to 99%, while reducing the energy consumption by up to 18%. Hipster is also e ective in learning and adapting automatically to speci c requirements of new incoming workloads just enough to meet the QoS and optimize resource consumption. In this work, we extend our previous work in several ways. First, we present an analysis of the size of the reward lookup table and an optimization for the table to improve the scalability of our reinforcement learning mechanism. Second, we demonstrate Hipster's capability to adapt to changes in the latency-critical application at runtime and still satisfy QoS guarantees of the new incoming applications. Lastly, we present a deployment methodology for setting up new applications managed by Hipster's runtime system.