Justice: A Deadline-Aware, Fair-Share Resource Allocator for Implementing Multi-Analytics

Dimopoulos, Stratos; Krintz, Chandra; Wolski, Rich

doi:10.1109/cluster.2017.52

Cited by 16 publications

(7 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We present our vision for a distributed two-level scheduler based on Justice [5] suitable for multi-cloudlet environments with resource offloading to remote public clouds. We discuss our motivation and analyze the new feature set that Justice must support to preserve fairness and satisfy job latency requirements in the context of a multi-cloudlet and public-cloud hybrid environments and MCC workloads.…”

Section: Discussionmentioning

confidence: 99%

“…Justice [5] is an allocator designed for the resource-constrained settings of a cloudlet with the dual goal of preserving fairness and satisfying job deadlines for multi-analytics batch workloads. Justice does so by using an adaptive prediction technique based on historical job execution times to estimate the minimum number of CPUs a job requires to meet its deadline "just-in-time".…”

Section: Background and Related Workmentioning

confidence: 99%

“…To allow low-latency processing, we have designed a predictive resource allocation mechanism based on admission control that can also adapt to changing cloud conditions in order to satisfy application deadlines and preserve fairness in resource-constrained cloud environments similar to those found in cloudlets. This allocator called Justice [5] is an application-agnostic framework for processing a variety of data analytics workloads. It employs a coarsegrained predictive mechanism that imposes very little overhead, making it ideal for resource-limited environments.…”

mentioning

confidence: 99%

See 2 more Smart Citations

Towards distributed, fair, deadline-driven resource allocation for cloudlets

Dimopoulos

Krintz

Wolski

2019

Proceedings of the 4th Workshop on Middleware for Edge Clouds &Amp; Cloudlets - MECC '19

Self Cite

View full text Add to dashboard Cite

In this paper we present our vision for a two-level, distributed resource allocator that preserves fairness and satisfies deadlines of low latency workloads in a multi-cloudlet environment with offloading support. We analyze the opportunities and challenges that offloading and the multi-cloud environment impose and we suggest the changes required to a fair-preserving and deadline-driven resource allocator originally designed for resource-constrained environments. CCS CONCEPTS • Computer systems organization → Cloud computing.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Background and Related Workmentioning

confidence: 99%

mentioning

confidence: 99%

See 1 more Smart Citation

Towards distributed, fair, deadline-driven resource allocation for cloudlets

Dimopoulos

Krintz

Wolski

2019

Proceedings of the 4th Workshop on Middleware for Edge Clouds &Amp; Cloudlets - MECC '19

Self Cite

View full text Add to dashboard Cite

show abstract

“…A live video streaming service was used to demonstrate the performance of the system. In [33] a resource allocator named Justice is proposed for cluster managers. The Justice uses deadline information of a job and historical job execution times to improve deadline satisfaction and fairness.…”

Section: Middleware Platforms For Distributed Computing Systemsmentioning

confidence: 99%

“…A compariso vided in Table 3. project "mobile through a mobile The key differen cloud system we [33] machine learning-based algorithms and schemes are also not de-AN infrastructure, and therefore, they do not utilize a vast amount etwork.…”

Section: Ad Hoc Cloudmentioning

confidence: 99%

Design of a Machine Learning-Based Intelligent Middleware Platform for a Heterogeneous Private Edge Cloud System

Shah

2021

Sensors

View full text Add to dashboard Cite

Recent advances in mobile technologies have facilitated the development of a new class of smart city and fifth-generation (5G) network applications. These applications have diverse requirements, such as low latencies, high data rates, significant amounts of computing and storage resources, and access to sensors and actuators. A heterogeneous private edge cloud system was proposed to address the requirements of these applications. The proposed heterogeneous private edge cloud system is characterized by a complex and dynamic multilayer network and computing infrastructure. Efficient management and utilization of this infrastructure may increase data rates and reduce data latency, data privacy risks, and traffic to the core Internet network. A novel intelligent middleware platform is proposed in the current study to manage and utilize heterogeneous private edge cloud infrastructure efficiently. The proposed platform aims to provide computing, data collection, and data storage services to support emerging resource-intensive and non-resource-intensive smart city and 5G network applications. It aims to leverage regression analysis and reinforcement learning methods to solve the problem of efficiently allocating heterogeneous resources to application tasks. This platform adopts parallel transmission techniques, dynamic interface allocation techniques, and machine learning-based algorithms in a dynamic multilayer network infrastructure to improve network and application performance. Moreover, it uses container and device virtualization technologies to address problems related to heterogeneous hardware and execution environments.

show abstract

Optimizing Cost and Maximizing Profit for Multi-Cloud-Based Big Data Computing by Deadline-Aware Optimize Resource Allocation

Manekar¹,

Pradeepini²

2020

Recent Studies on Computational Intelligence

View full text Add to dashboard Cite

Justice: A Deadline-Aware, Fair-Share Resource Allocator for Implementing Multi-Analytics

Cited by 16 publications

References 35 publications

Towards distributed, fair, deadline-driven resource allocation for cloudlets

Towards distributed, fair, deadline-driven resource allocation for cloudlets

Design of a Machine Learning-Based Intelligent Middleware Platform for a Heterogeneous Private Edge Cloud System

Optimizing Cost and Maximizing Profit for Multi-Cloud-Based Big Data Computing by Deadline-Aware Optimize Resource Allocation

Contact Info

Product

Resources

About