Autonomic performance-per-watt management (APM) of cloud resources and services

Fargo, Farah; Tunc, Cihan; Al-Nashif, Youssif; Hariri, Salim

doi:10.1145/2494621.2494624

Cited by 10 publications

(11 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In order to show the effectiveness of the proposed methods, we compare our work with three different methods: (1) static resource allocation (where the resources are pre-defined using a pessimistic approach), (2) adaptive core frequency scaling method discussed in [7] as a state-of-the-art heuristic based power controller (that prioritizes power limit over high performance and this approach is commonly used in datacenters for the power control of servers [7]), and (3) the previous approach from [6].…”

Section: G Comparison Methodsmentioning

confidence: 99%

“…In addition to the aforementioned works, the closest approach for performance management to our works are [30], [31], and [6]. In [30], the proposed method is scaling up/down the VM resources on runtime based on the utilization ratios (i.e.…”

Section: F Overall Summarymentioning

confidence: 99%

“…However, in their work, they only consider consolidation and the numbers of CPUs for the VMs and they exclude memory amount and the core frequencies. Among them [6] is the first study with multiple resource management for power and performance.…”

Section: F Overall Summarymentioning

confidence: 99%

“…These features enable the management system to predict trends in the dynamic cloud resource requirements of the workload and to proactively configure the cloud resources to meet these requirements ahead of time. For further detail about AppFlow and how it can be used to characterize and predict workloads, please refer to [16,32,6].…”

Section: A Appflow: a Data Structure For Autonomic Managementmentioning

confidence: 99%

“…In a previous work ( [6]), performance-per-Watt metric was used for scaling up/down the virtual machines' resources during runtime. This method achieved reduction in power consumption with little performance loss (less than 2%) it reduced the overall power consumption up to 84% compared to the static method and up to 33% compared to the frequency adaptive method for online bidding workloads using RUBiS benchmark.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Autonomic Workload and Resources Management of Cloud Computing Services

Fargo

Tunc

Al-Nashif

et al. 2014

2014 International Conference on Cloud and Autonomic Computing

Self Cite

View full text Add to dashboard Cite

The power consumption of data centers and cloud systems have increased almost three times between 2007 and 2012. Over-provisioning techniques are typically used for meeting the peak workloads. In this paper we present an autonomic power and performance management method for cloud systems in order to dynamically match the application requirements with "just-enough" system resources at runtime that lead to significant power reduction while meeting the quality of service requirements of the cloud applications. Our solution offers the following capabilities: 1) real-time monitoring of the cloud resources and workload behavior running on virtual machines (VMs); 2) determine the current operating point of both workloads and the VMs running these workloads; 3) characterize workload behavior and predict the next operating point for the VMs; 4) dynamically manage the VM resources (scaling up and down the number of cores, CPU frequency, and memory amount) at run time; and 5) assign available cloud resources that can guarantee optimal power consumption without sacrificing the QoS requirements of cloud workloads. We validate the performance of our approach using the RUBiS benchmark, an auction model emulating eBay transactions that generates a wide range of workloads (such as browsing and bidding with different number of clients). Our experimental results show that our approach can lead to reduction in power consumption up to 87% when compared to the static resource allocation strategy, 72% compared to adaptive frequency scaling strategy and 66% compared to a similar multi-resource management strategy.

show abstract

Section: G Comparison Methodsmentioning

confidence: 99%

Section: F Overall Summarymentioning

confidence: 99%

Section: F Overall Summarymentioning

confidence: 99%

Section: A Appflow: a Data Structure For Autonomic Managementmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Autonomic Workload and Resources Management of Cloud Computing Services

Fargo

Tunc

Al-Nashif

et al. 2014

2014 International Conference on Cloud and Autonomic Computing

Self Cite

View full text Add to dashboard Cite

show abstract

Joint‐analysis of performance and energy consumption when enabling cloud elasticity for synchronous HPC applications

Righi

Costa

Rodrigues

et al. 2015

Concurrency and Computation

View full text Add to dashboard Cite

Summary A key characteristic of cloud computing is elasticity, automatically adjusting system resources to an application's workload. Both reactive and horizontal approaches represent traditional means to offer this capability, in which rule‐condition‐action statements and upper and lower thresholds occur to instantiate or consolidate compute nodes and virtual machines. Although elasticity can be beneficial for many HPC (high‐performance computing) scenarios, it also imposes significant challenges in the development of applications. In addition to issues related to how we can incorporate this new feature in such applications, there is a problem associated with the performance and resource pair and, consequently, with energy consumption. Further exploring this last difficulty, we must be capable of analyzing elasticity effectiveness as a function of employed thresholds with clear metrics to compare elastic and non‐elastic executions properly. In this context, this article explores elasticity metrics in two ways: (i) the use of a cost function that combines application time with different energy models; (ii) the extension of speedup and efficiency metrics, commonly used to evaluate parallel systems, to cover cloud elasticity. To accomplish (i) and (ii), we developed an elasticity model known as AutoElastic, which reorganizes resources automatically across synchronous parallel applications. The results, obtained with the AutoElastic prototype using the OpenNebula middleware, are encouraging. Considering a CPU‐bound application, an upper threshold close to 70% was the best option for obtaining good performance with a non‐prohibitive elasticity cost. In addition, the value of 90% for this threshold was the best option when we plan an efficiency‐driven execution. Copyright © 2015 John Wiley & Sons, Ltd.

show abstract