Non-linear analysis of bursty workloads using dual metrics for better cloud resource management

Balaji, Mahesh; Cherukuri, Aswani Kumar; Rao, G Sudhakara

doi:10.1007/s12652-019-01183-8

Cited by 9 publications

(4 citation statements)

References 75 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…While this approach is effective in ensuring the timely processing of important tasks, it may neglect the needs of lower-priority tasks, leading to potential underutilization of resources. [12] Additionally, defining accurate and fair priorities can be challenging, and the system may not adapt well to the evolving priorities of AI workloads.…”

Section: 1c Priority-based Allocationmentioning

confidence: 99%

Efficient Workload Allocation and Scheduling Strategies for AI-Intensive Tasks in Cloud Infrastructures.

Ahmad Fawad

2023

pst

View full text Add to dashboard Cite

- The rapid proliferation of Artificial Intelligence (AI) applications has underscored the need for advanced cloud infrastructures capable of efficiently managing AI-intensive workloads. This paper delves into the intricacies of workload allocation and scheduling in the context of cloud environments, specifically focusing on the challenges posed by AI-intensive tasks. Our research endeavors to scrutinize existing strategies, discern their limitations, and proffer innovative approaches tailored to optimize the allocation and scheduling of AI workloads within cloud infrastructures. In elucidating the challenges, we pinpoint resource heterogeneity, dynamic workload characteristics, and scalability as the crux of the issues confronting AI-intensive workload management. The diverse computational demands of AI workloads make it challenging to allocate resources optimally, while the dynamic nature of these tasks necessitates adaptive strategies to accommodate varying computational requirements over time. [1] Additionally, as AI models and datasets burgeon in complexity and size, ensuring scalability becomes paramount for sustaining performance in cloud environments. Our literature review encompasses an examination of both traditional and state-of-the-art workload allocation strategies, shedding light on their respective strengths and shortcomings. We also delve into scheduling techniques employed for managing AI-intensive tasks, providing a comprehensive overview of the existing landscape. To address these challenges, we propose a novel framework centered around dynamic resource provisioning, machine learning-based scheduling, and efficient task migration strategies. The framework aims to adaptively allocate resources based on the evolving nature of AI workloads, leveraging machine learning algorithms to predict workload characteristics and employing efficient task migration to handle workload fluctuations. The paper concludes with an experimental evaluation of the proposed strategies, conducted in a simulated environment using diverse datasets. Key performance metrics, such as throughput, latency, and resource utilization, are employed to assess the effectiveness of our strategies compared to existing approaches. By offering insights into the efficient management of AI-intensive workloads in cloud infrastructures, this research contributes to the ongoing efforts to enhance the scalability and performance of cloud environments in the face of burgeoning AI applications.

show abstract

Section: 1c Priority-based Allocationmentioning

confidence: 99%

Efficient Workload Allocation and Scheduling Strategies for AI-Intensive Tasks in Cloud Infrastructures.

Ahmad Fawad

2023

pst

View full text Add to dashboard Cite

show abstract

“…A few efforts are made to measure the burstiness in application workloads. For example, Balaji et al [21] use a combination of Hurst Exponent and Sample Entropy methods to detect burst patterns using offline workload traces. Zhang et al [42] use a two-state (ON/OFF) Markov chain model to detect the burst in the given workload.…”

Section: Related Workmentioning

confidence: 99%

Online Workload Burst Detection for Efficient Predictive Autoscaling of Applications

et al. 2020

View full text Add to dashboard Cite

Autoscaling methods are employed to ensure the scalability of cloud-hosted applications. The public-facing applications are prone to receive sudden workload bursts, and the existing autoscaling methods do not handle the bursty workloads gracefully. It is challenging to detect the burst online from the incoming dynamic workload traffic, and then identifying appropriate resources to address the burst without overprovisioning is even harder. In this paper, we address this challenge by investigating the appropriate method for online burst detection and then proposed a novel predictive autoscaling method to use burst detection for satisfying specific response time requirements. We compared the proposed method with multiple state-of-the-art baseline autoscaling methods under multiple realistic and synthetic bursty workloads for a benchmark application. Our experimental results show a 60.8% average decrease in response time violations as compared to the baseline method. INDEX TERMS Autoscaling, predictive, SLO violations, response time, workload, burstiness, online burst detection.

show abstract

“…The authors, as part of their earlier studies, had introduced a predictive resource management framework (PRMF) with the objective of provisioning resources using a predictive approach. 5,8 PRMF consists of statistical libraries and a three-stage workflow. The first of the three is the feature selection stage.…”

Section: Categorize User-request Based On Wtmentioning

confidence: 99%

“…Enterprise workloads are complex and nonstationary due to large, concurrent, and dependent applications that get executed in parallel or sequence. Simultaneous requests from such applications in a short time‐period result in burst/spike, leading to resource management challenges 7‐9 . An efficient Cloud resource management framework needs to classify the user request into workload type (WT) to offset this uncertainty.…”

Section: Introductionmentioning

confidence: 99%

Context‐aware resource management and alternative pricing model to improve enterprise cloud adoption

Balaji

Kumar

2020

Concurrency and Computation

Self Cite

View full text Add to dashboard Cite

Existing Cloud resource management approaches favor Cloud service providers because of their data-center origin. The onus of efficiently managing the resources is on the end-user. Present provisioning/deprovisioning methods and the pricing model could be modified to ensure that the end-user reaps the benefits. The current study proposes a context-aware Cloud resource management approach based on a workload-type provisioning and a just-in-time deprovisioning design. Furthermore, it also recommends an alternative pricing model. The design intends to provision/deprovision individual resources like CPU, memory, and so on, according to the workload-type rather than instance-type as done in existing Cloud resource management approaches. The proposed approach was commissioned on a private lab infrastructure and several experiments were conducted. It was observed that the proposed approach was able to detect sample workload types with an average sensitivity of 82% and specificity of 76%. The average cost of the proposed approach was found to be 20% less than the present approach. K E Y W O R D S alternative cloud pricing, cloud resource management, cloud scaling, just-in-time deprovisioning 1 INTRODUCTION Cloud computing is an evolving paradigm and end-users could benefit more from this model if the supporting technologies like resource management and pricing provide more options. Existing resource management approaches are either rule-based and reactive (subsequently referred to as the traditional approach) or predictive. 1-4 The authors, in their earlier study, observed that provisioning was better using the predictive approach than the traditional approach. 5,6 Enterprise workloads are complex and nonstationary due to large, concurrent, and dependent applications that get executed in parallel or sequence. Simultaneous requests from such applications in a short time-period result in burst/spike, leading to resource management challenges. 7-9 An efficient Cloud resource management framework needs to classify the user request into workload type (WT) to offset this uncertainty. The WT of a user-request prescribes the resources required for fulfilling the demand. For example, a compute-intensive on-off WT would rather require additional CPU than storage, while a read-write-intensive exponential WT would require more disk IOPS than CPU. Earlier studies report a measure of self-similarity as a proficient approach for identifying anomalies that could be leveraged for WT classification. 10-12 It must be noted that existing Cloud resource management approaches do not include WT while provisioning resources. Currently, Cloud resources such as CPU, memory, disk, and network are grouped into logical units called instance-types (virtual machines) for provisioning/deprovisioning. Although the demand for individual resources like CPU, memory, and so on, is a frequent occurrence in enterprise Cloud applications, the only choice for the end-user is to make use of instance-types. An instance-type-based approach would make it is easier f...

show abstract

Non-linear analysis of bursty workloads using dual metrics for better cloud resource management

Cited by 9 publications

References 75 publications

Efficient Workload Allocation and Scheduling Strategies for AI-Intensive Tasks in Cloud Infrastructures.

Efficient Workload Allocation and Scheduling Strategies for AI-Intensive Tasks in Cloud Infrastructures.

Online Workload Burst Detection for Efficient Predictive Autoscaling of Applications

Context‐aware resource management and alternative pricing model to improve enterprise cloud adoption

Contact Info

Product

Resources

About