Dynamic resource demand prediction and allocation in multi‐tenant service clouds

Verma, Manish; Gangadharan, G. R.; Narendra, Nanjangud C.; Ravi, Vadlamani; Inamdar, Vidyadhar; Ramachandran, Lakshmi; Calheiros, Rodrigo N.; Buyya, Rajkumar

doi:10.1002/cpe.3767

Cited by 55 publications

(15 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The ARIMA based prediction model was considered to improve the utilization and response time for users. A framework for allocation in cloud was presented by Verma et al in [16]. Here, based on the dynamic nature of resource requirements, service tenants were classified and its resource prediction was prioritized to minimize the prediction time.…”

Section: Related Workmentioning

confidence: 99%

Cloud capacity planning and HSI based optimal resource provisioning

Sadashiv¹,

Kumar²,

Goudar³

2017

2017 Second International Conference on Electrical, Computer and Communication Technologies (ICECCT)

View full text Add to dashboard Cite

Cloud service providers offer spot instances through highest bidding plans that are at a very economical price compared to other pricing plans, namely on-demand and reservation. The usage of spot instance enables utilization of idle resources and provide service for cost sensitive tasks. However, this approach introduces the problem of cloud capacity allocation to different pricing plans that will have impact on the task completion time. To address these issues and improve the providers revenue, in this paper a capacity planning has been carried out based on the prediction of resource requirements for each of the different resource pricing pools. The paper also presents a solution to overcome the burden faced by the service provider due to the free issue of last hour at the time of out-of-bid situation. Simulation carried out based on capacity planning along with hybrid spot instance using Amazon EC2's price show that the resource utilization is improved across the different resource pricing pools with increased number of task completion and improved provider's revenue.

show abstract

Section: Related Workmentioning

confidence: 99%

Cloud capacity planning and HSI based optimal resource provisioning

Sadashiv¹,

Kumar²,

Goudar³

2017

2017 Second International Conference on Electrical, Computer and Communication Technologies (ICECCT)

View full text Add to dashboard Cite

show abstract

“…In order to manage the mutual interference between the device-to-device links and the cellular links, effective resource allocation mechanisms are needed. The resources assigned to satisfy the alteration in demands for services are adapted by dynamic resource assignment techniques [14].…”

Section: Introductionmentioning

confidence: 99%

Dynamic Resource Allocation Model for Distribution Operations Using SDN

Goudarzi

Anisi

Ahmadi

et al. 2021

IEEE Internet Things J.

View full text Add to dashboard Cite

In vehicular ad-hoc networks, autonomous vehicles generate a large amount of data prior to support in-vehicle applications. So, a big storage and high computation platform is needed. On the other hand, the computation for vehicular networks at the cloud platform requires low latency. Applying edge computation (EC) as a new computing paradigm has potentials to provide computation services while reducing the latency and improving the total utility. We propose a three-tier EC framework to set the elastic calculating processing capacity and dynamic route calculation to suitable edge servers for real-time vehicle monitoring. This framework includes the cloud computation layer, EC layer, and device layer. The formulation of resource allocation approach is similar to an optimization problem. We design a new reinforcement learning (RL) algorithm to deal with resource allocation problem assisted by cloud computation. By integration of EC and software defined networking (SDN), this study provides a new software defined networking edge (SDNE) framework for resource assignment in vehicular networks. The novelty of this work is to design a multi-agent RL-based approach using experience reply. The proposed algorithm stores the users' communication information and the network tracks' state in realtime. The results of simulation with various system factors are presented to display the efficiency of the suggested framework. We present results with a real-world case study.

show abstract

“…Time series analysis has been widely used to implement auto-scaling mechanisms for applications that exhibit some kind of temporal patterns. Most of these proposals (e.g., [7][8][9][10]) use linear statistical methods for time-series forecasting, mainly based on Box and Jenkins [11] autoregressive models (e.g., AR, ARMA, ARIMA, or ARMAX) for predicting service metrics (e.g., the server load) from historical observations. However, as these service metrics can exhibit non-linear patterns, some key features of the input data may not be properly captured by these linear models.…”

Section: Introductionmentioning

confidence: 99%

Efficient resource provisioning for elastic Cloud services based on machine learning techniques

et al. 2019

View full text Add to dashboard Cite

Automated resource provisioning techniques enable the implementation of elastic services, by adapting the available resources to the service demand. This is essential for reducing power consumption and guaranteeing QoS and SLA fulfillment, especially for those services with strict QoS requirements in terms of latency or response time, such as web servers with high traffic load, data stream processing, or real-time big data analytics. Elasticity is often implemented in cloud platforms and virtualized data-centers by means of auto-scaling mechanisms. These make automated resource provisioning decisions based on the value of specific infrastructure and/or service performance metrics. This paper presents and evaluates a novel predictive auto-scaling mechanism based on machine learning techniques for time series forecasting and queuing theory. The new mechanism aims to accurately predict the processing load of a distributed server and estimate the appropriate number of resources that must be provisioned in order to optimize the service response time and fulfill the SLA contracted by the user, while attenuating resource over-provisioning in order to reduce energy consumption and infrastructure costs. The results show that the proposed model obtains a better forecasting accuracy than other classical models, and makes a resource allocation closer to the optimal case.

show abstract

Dynamic resource demand prediction and allocation in multi‐tenant service clouds

Cited by 55 publications

References 31 publications

Cloud capacity planning and HSI based optimal resource provisioning

Cloud capacity planning and HSI based optimal resource provisioning

Dynamic Resource Allocation Model for Distribution Operations Using SDN

Efficient resource provisioning for elastic Cloud services based on machine learning techniques

Contact Info

Product

Resources

About