WASP: Workload Adaptive Energy-Latency Optimization in Server Farms Using Server Low-Power States

Yao, Fan; Wu, Jingxin; Subramaniam, Suresh; Venkataramani, Guru

doi:10.1109/cloud.2017.30

Cited by 17 publications

(12 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…One common host-level power saving technique, DVFS, can lower the power consumption by running workloads more slowly. DVFS policies and mechanisms have become more sophisticated over time, starting from power throttling of a chip multi-processor [21], to finer-grain per-core management [37,39], to workload-and request-aware throttling decisions [9,10,20,23,31,45,46]. Newer approaches that can work in high-utilization environments generally require visibility into application-exported metrics.…”

Section: Why Consider Applications As Opaque Boxes?mentioning

confidence: 99%

“…Kanev et al [19] bring power management to the data center era by examining the effect of c-states and DVFS on the performance and power consumption of cloud workloads. CARB [46], WASP [45], and DynSleep [10] achieve power savings through deep sleep c-states while retaining tail latency constraints. Pegasus [23] adjusts voltage and frequency at regular intervals in response to changes in the application-reported latency.…”

Section: Mechanismmentioning

confidence: 99%

See 1 more Smart Citation

Leveraging application classes to save power in highly-utilized data centers

Kaffes

Sbîrlea

Lin

et al. 2020

Proceedings of the 11th ACM Symposium on Cloud Computing

View full text Add to dashboard Cite

Data center energy consumption has become an increasingly significant contributor both to greenhouse emissions and costs. To increase utilization of individual hosts and improve efficiency, most modern data centers co-locate workloads belonging to different application classes, some being latency-sensitive (LS) and others best-effort (BE) which are more tolerant to performance variation. It is therefore necessary to design mechanisms that reduce power consumption even in the resulting high-utilization environment, while preserving LS task performance. Moreover, the abundance of different workloads and the security implications of public cloud make mechanisms that rely on extensive knowledge of workload characteristics or on application-exported metrics challenging to deploy. We present PACT, Per Application Class Turbo Controller, a system that leverages two novel mechanisms to reduce power consumption even in highly-utilized data centers. By treating applications like opaque boxes that do not need to provide application-specific performance signals, the first mechanism, Turbo Control, reduces power consumption by decreasing the operating frequency and throttling only BE tasks, without affecting performance-sensitive LS tasks. We identify the shortcomings of Turbo Control and increase its effectiveness by introducing CPUJailing, a mechanism that allocates different sets of cores to LS and BE applications. We deploy PACT (Turbo Control + CPUJailing) in production * Kostis Kaffes was an intern at Google during this work. † Christos Kozyrakis was partly at Google during this work.

show abstract

Section: Why Consider Applications As Opaque Boxes?mentioning

confidence: 99%

Section: Mechanismmentioning

confidence: 99%

Leveraging application classes to save power in highly-utilized data centers

Kaffes

Sbîrlea

Lin

et al. 2020

Proceedings of the 11th ACM Symposium on Cloud Computing

View full text Add to dashboard Cite

show abstract

“…Right-sizing the energy budgets of compute and cooling, while maintaining the Quality of Service (QoS) requirements in data centers, is a challenging problem [20]. Prior works, that optimize system energy under QoS constraints of workloads, typically consolidate the workload onto the fewest number of servers so that other servers can remain in sleep (inactive) states [7,9,10,19]. While such a strategy is effective to reduce CPU power, it can inadvertently raise the thermal profile of active servers, resulting in increased power consumption of its fans.…”

Section: Introductionmentioning

confidence: 99%

“…To reduce energy consumption of servers in data centers, prior works have proposed the use of low power states [7,10]. However, higher job latency can be caused by placing servers into low power states, resulting in the failure to meet the Quality of Service (QoS) requirements [19]. Job service latency can be high due to (1) queuing effects -for example, this will happen if there are not enough servers to handle a burst of arriving jobs; (2) having to wake up from low power states.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

SpinSmart

Tian

Vishwanath

Venkataramani

et al. 2020

Proceedings of the Eleventh ACM International Conference on Future Energy Systems

Self Cite

View full text Add to dashboard Cite

Cost of data centers has risen sharply in the past few years. Today, it represents about 3% of total US energy consumption with projections to increase further in the coming years. In this paper, we focus on the server infrastructure and observe that workload consolidation techniques, which maximize power efficiency of server systems, do not automatically optimize the overall system power efficiency especially when compute engines and the corresponding on-board cooling systems are considered holistically. We design SpinSmart, a framework that explores optimal server fan speeds to minimize the overall system energy consumption. We explore core capping strategies that estimate the desired number of CPU cores to be used at any given time to minimize combined CPU+fan power. Our experimental results show that we are able to achieve 1) energy savings of up to 10% of total energy and 80% of cooling energy when compared to workload consolidation without core capping strategy; 2) cooling energy savings up to 42% when compared to the strategy that randomly assigns jobs to all the servers and cores. CCS CONCEPTS• Hardware → Temperature control; Temperature optimization.

show abstract

Online Power-Aware Deployment and Load Distribution Optimization for Application Server Clusters

et al. 2019

View full text Add to dashboard Cite

The energy conservation of application server clusters is a pressing problem. In this paper, we propose an online power-aware deployment and load distribution optimization strategy for application server clusters, whose objective is to minimize the cluster's power while ensuring that the server's CPU utilization is not higher than a preset value. The strategy includes two schemes: a mixed integer linear programming (MILP)-based scheme and a mixed integer non-linear programming (MINLP)-based scheme. The former formulates the cluster optimization problem as a MILP problem and adopts a toolkit to solve it. When the cluster scale is small, it can find the global optimal solution quickly. So, the MILP-based scheme is applicable to small-scale clusters. In the MINLP-based scheme, we first formulate the cluster optimization problem as a non-linear programming problem, and then design a method to reduce the number of variables and reformulate it as an MINLP problem. We finally propose an efficient solution method based on the flower pollination algorithm. Due to the small number of variables and the high solution efficiency, the solution method can quickly obtain a high-quality solution, so the MINLP-based scheme can be applied to largescale clusters. The experimental results demonstrate the effectiveness of our strategy.

show abstract

WASP: Workload Adaptive Energy-Latency Optimization in Server Farms Using Server Low-Power States

Cited by 17 publications

References 20 publications

Leveraging application classes to save power in highly-utilized data centers

Leveraging application classes to save power in highly-utilized data centers

SpinSmart

Online Power-Aware Deployment and Load Distribution Optimization for Application Server Clusters

Contact Info

Product

Resources

About