A Host-Agnostic, Supervised Machine Learning Approach to Automated Overload Detection in Virtual Machine Workloads

Dow, Eli M.; Matthews, Jeanna

doi:10.1109/smartcloud.2017.10

Cited by 3 publications

(9 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…detailed study of related work in the domain of VM consolidation. A very few authors have focused specifically on host overload detection in their research, such as Beloglazov et al, [11][12][13] Dow et al, 14,15 Melhem et al, 16 Tseng et al, 17 Wang et al, 18 and Heyang et al 19 Table 1 represents the objective, corresponding methodology, performance evaluation parameters, type of workload, and simulator used by various authors. Beloglazov et al 22,23 presented the VM placement algorithm-modified Best Fit Decreasing (BFD) algorithm.…”

Section: Background Studymentioning

confidence: 99%

“…This is due to the fact that the input workload used during simulation is not seasonal or trendy. Table 5 shows the nature of input workload as random, with average CPU utilization in the range [09-13] and standard deviation in between [12][13][14][15][16][17][18]. For each input workload, energy consumed by EWMA2 is at least equal or more than Figure 5 shows the SLA violation comparison of proposed variants of EWMA.…”

Section: Comparative Analysis Of Proposed Algorithmsmentioning

confidence: 99%

“…Dow et al 14,15 have worked specifically on host overload detection. They have used a supervised machine learning algorithm, decision tree classifier, to classify a host as overloaded.…”

Section: Background Studymentioning

confidence: 99%

“…Later, Beloglazov et al 13 and Melhem et al 16 proposed Markov chain model-based policy for host overload detection. Dow et al 14,15 have used a supervised machine learning algorithm, decision tree classifier for predicting overloaded hosts on KVM hypervisor. Metaheuristic has also been a choice for optimal allocation of VMs on a given set of hosts.…”

mentioning

confidence: 99%

“…VM consolidation is further explained in Section 3. Among the three above-mentioned subprocesses of VM consolidation, a very few authors have focused specifically on host overload detection in their research, such as Beloglazov et al, [11][12][13] Dow et al, 14,15 Melhem et al, 16 Tseng et al, 17 Wang et al, 18 and Heyang et al 19 Beloglazov et al 11,12 proposed state-of-the-art algorithms based on local regression (LR), robust local regression (LRR), median absolute deviation (MAD), interquartile range (IQR), and upper threshold (THR). Among the above-proposed algorithms, LR proved to be the best performing.…”

mentioning

confidence: 99%

See 4 more Smart Citations

An efficient host overload detection algorithm for cloud data center based on exponential weighted moving average

Kulshrestha¹,

Patel

2021

Int J Communication

View full text Add to dashboard Cite

The major aim of data center (DC) is to optimize the three parameters, viz., energy consumption, resource utilization, and quality of service (QoS). Our objective is to balance these three parameters by managing the virtual machine (VM) efficiently. Therefore, this paper presents a holistic view of VM consolidation and explains its constituent subprocesses: (i) host overload detection, (ii) VM selection for migration from the overloaded host, and (iii) VM placement for provisioning selected VMs on a new set of hosts. Later, this paper attempts to optimize the parameters mentioned above by proposing a host overload detection algorithm based on exponential weighted moving average (EWMA) and its variants. The performance of the proposed EWMA is compared with the stateof-the-art algorithms based on (i) local regression, (ii) median absolute deviation, and (iii) interquartile range. The efficacy of the proposed EWMA is evaluated in combination with four different existing VM selection policies: (i) minimum migration time, (ii) minimum utilization, (iii) maximum correlation, and (iv) random selection. The evaluation metrics comprise energy consumption, VM migration count from the overloaded hosts representing the resource utilization, service level agreement (SLA) violations describing the QoS, and the average execution time. Simulation experiments are carried out on the CloudSim simulator using PlanetLab real cloud trace as workload. The proposed EWMA consumes 18.33% less energy, causes 47.81% less VM migration, makes 9.91% less SLA violation, and takes 43.44% less average-time for the execution of the entire workload as compared to state-of-the-art algorithms.

show abstract

Section: Background Studymentioning

confidence: 99%

Section: Comparative Analysis Of Proposed Algorithmsmentioning

confidence: 99%

“…Dow et al 14,15 have worked specifically on host overload detection. They have used a supervised machine learning algorithm, decision tree classifier, to classify a host as overloaded.…”

Section: Background Studymentioning

confidence: 99%

mentioning

confidence: 99%

mentioning

confidence: 99%

See 3 more Smart Citations

An efficient host overload detection algorithm for cloud data center based on exponential weighted moving average

Kulshrestha¹,

Patel

2021

Int J Communication

View full text Add to dashboard Cite

show abstract

Anomaly Detection for Black Box Services in Edge Clouds Using Packet Size Distribution

Wallschläger¹,

Gulenko²,

Schmidt³

et al. 2018

2018 IEEE 7th International Conference on Cloud Networking (CloudNet)

View full text Add to dashboard Cite

Cloud Host Selection using Iterative Particle-Swarm Optimization for Dynamic Container Consolidation

Ravikumar

Begum²,

Kumar³

et al. 2022

IJRITCC

View full text Add to dashboard Cite

A significant portion of the energy consumption in cloud data centres can be attributed to the inefficient utilization of available resources due to the lack of dynamic resource allocation techniques such as virtual machine migration and workload consolidation strategies to better optimize the utilization of resources. We present a new method for optimizing cloud data centre management by combining virtual machine migration with workload consolidation. Our proposed Energy Efficient Particle Swarm Optimization (EE-PSO) algorithm to improve resource utilization and reduce energy consumption. We carried out experimental evaluations with the Container CloudSim toolkit to demonstrate the effectiveness of the proposed EE-PSO algorithm in terms of energy consumption, quality of service guarantees, the number of newly created VMs, and container migrations.

show abstract

A Host-Agnostic, Supervised Machine Learning Approach to Automated Overload Detection in Virtual Machine Workloads

Cited by 3 publications

References 12 publications

An efficient host overload detection algorithm for cloud data center based on exponential weighted moving average

An efficient host overload detection algorithm for cloud data center based on exponential weighted moving average

Anomaly Detection for Black Box Services in Edge Clouds Using Packet Size Distribution

Cloud Host Selection using Iterative Particle-Swarm Optimization for Dynamic Container Consolidation

Contact Info

Product

Resources

About