Heracles

Lo, David; Cheng, Liqun; Govindaraju, Rama; Ranganathan, Parthasarathy; Kozyrakis, Christos

doi:10.1145/2872887.2749475

Cited by 65 publications

(25 citation statements)

References 74 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Corroborating the ndings of prior work [41], we observe that collocating both latency-critical and batch workloads degrades QoS at higher loads due to shared resource contention. If the reward mechanism is not aware of such collocations, it may make decisions that violate QoS for the latency-critical workload and/or reduce the throughput of the batch workloads.…”

Section: Reward Mechanism For Hipstercosupporting

confidence: 80%

“…3 While regression models may have lower overhead than RL approaches, we show in our experiments that Hipster's overhead is still <0.2% (see Section 3.10); this is similar to the overhead incurred by prior work [41,56] without using RL techniques.…”

Section: Machine Learning To Design E Icient Systemsmentioning

confidence: 55%

“…Similarly, Facebook consistently sees diurnal load variations between 10% and 95% of maximum capacity, across multiple server clusters [4,8]. The periods of low server utilization provide opportunity to reduce data center energy consumption [17,18,41,56].…”

Section: Motivationmentioning

confidence: 99%

“…We evaluate Hipster using two latency-critical applications, Memcached and Web-Search, which have distinct characteristics and impact on shared resources [41]. Memcached [25] is an open source implementation of an in-memory key-value store for data caching used in many services from Twitter, Facebook and Google [4,51].…”

Section: Evaluation 41 Cloud Benchmarksmentioning

confidence: 99%

“…Heracles [41] uses a feedback controller that exploits collocation of latency-critical and batch workloads while increasing the resource e ciency of CPU, memory and network as long as QoS target is met. However, this work explores modern Intel architectures with cache allocation technology (CAT) and DRAM bandwidth monitor, which are available from Broadwell processors released after 2015. Pegasus [40] achieves high CPU energy proportionality for low latency workloads using ne-grained DVFS techniques.…”

Section: Related Workmentioning

confidence: 99%

See 4 more Smart Citations

The Hipster Approach for Improving Cloud System Efficiency

Nishtala

Carpenter

Petrucci

et al. 2017

ACM Trans. Comput. Syst.

View full text Add to dashboard Cite

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.

show abstract

Section: Reward Mechanism For Hipstercosupporting

confidence: 80%

Section: Machine Learning To Design E Icient Systemsmentioning

confidence: 55%

Section: Motivationmentioning

confidence: 99%

Section: Evaluation 41 Cloud Benchmarksmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

See 3 more Smart Citations

The Hipster Approach for Improving Cloud System Efficiency

Nishtala

Carpenter

Petrucci

et al. 2017

ACM Trans. Comput. Syst.

View full text Add to dashboard Cite

show abstract

Niyama: Node scheduling for cloud workloads with resource isolation

Thiyyakat

Kalambur

Sitaram³

2022

Concurrency and Computation

View full text Add to dashboard Cite

Cloud providers place tasks from multiple applications on the same resource pool to improve the resource utilization of the infrastructure. The consequent resource contention has an undesirable effect on latency-sensitive tasks. In this article, we present Niyama-a resource isolation approach that uses a modified version of deadline scheduling to protect latency-sensitive tasks from CPU bandwidth contention.Conventionally, deadline scheduling has been used to schedule real-time tasks with well-defined deadlines. Therefore, it cannot be used directly when the deadlines are unspecified. In Niyama, we estimate deadlines in intervals and secure bandwidth required for the interval, thereby ensuring optimal job response times. We compare our approach with cgroups: Linux's default resource isolation mechanism used in containers today. Our experiments show that Niyama reduces the average delay in tasks by 3×-20× when compared to cgroups. Since Linux's deadline scheduling policy is work-conserving in nature, there is a small drop in the server-level CPU utilization when Niyama is used naively. We demonstrate how the use of core reservation and oversubscription in the inter-node scheduler can be used to offset this drop; our experiments show a 1.3×-2.24× decrease in delay in job response time over cgroups while achieving high CPU utilization.

show abstract

Priority-Based Optimization of I/O Isolation for Hybrid Deployed Services

Zhang

Zhou

et al. 2019

Lecture Notes of the Institute for Computer Sciences, Social Informatics and Telecommunications Engineering

View full text Add to dashboard Cite

Heracles

Cited by 65 publications

References 74 publications

The Hipster Approach for Improving Cloud System Efficiency

The Hipster Approach for Improving Cloud System Efficiency

Niyama: Node scheduling for cloud workloads with resource isolation

Priority-Based Optimization of I/O Isolation for Hybrid Deployed Services

Contact Info

Product

Resources

About