A dynamic task scheduler tolerant to multiple hibernations in cloud environments
- PDF / 2,880,155 Bytes
- 23 Pages / 595.276 x 790.866 pts Page_size
- 52 Downloads / 165 Views
(0123456789().,-volV)(0123456789(). ,- volV)
A dynamic task scheduler tolerant to multiple hibernations in cloud environments Luan Teylo1
•
Luciana Arantes2 • Pierre Sens2 • Lucia M. A. Drummond1
Received: 6 January 2020 / Revised: 18 August 2020 / Accepted: 21 August 2020 Ó Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract Cloud platforms usually offer several types of Virtual Machines (VMs) with different guarantees in terms of availability and volatility, provisioning the same resource through multiple pricing models. For instance, in the Amazon EC2 cloud, the user pays per use for on-demand VMs while spot VMs are instances available at lower prices. However, a spot VM can be terminated or hibernated by EC2 at any moment. In this work, we propose the Hibernation-Aware Dynamic Scheduler (HADS) that schedules Bag-of-Tasks (BoT) applications with deadline constraints in both hibernation prone spots VMs and on-demand VMs. HADS aims at minimizing the monetary costs of executing BoT applications on Clouds ensuring that their deadlines are respected even in the presence of multiple hibernations. Results collected from experiments on Amazon EC2 VMs using synthetic applications and a NAS benchmark application show the effectiveness of HADS in terms of monetary costs when compared to on-demand VM only solutions. Keywords Cloud computing Dynamic BoT scheduling Temporal failures Spot VM hibernation Monetary cost minimization
1 Introduction In the past few years, cloud computing has emerged as an attractive option to run different classes of applications due to several advantages over other platforms, such as: (i) immediate access to computational resources, (ii) no upfront capital investments, and (iii) pay-per-use model. Some cloud providers offer several classes of Virtual Machines (VMs) with different guarantees in terms of availability and volatility, provisioning the same resource through multiple pricing models. Amazon EC2, for & Luan Teylo [email protected] Luciana Arantes [email protected] Pierre Sens [email protected] Lucia M. A. Drummond [email protected] 1
Instituto de Computac¸a¯o, Universidade Federal Fluminense, Nitero´i, Brazil
2
Sorbonne Universite´, CNRS, INRIA, LIP6, Paris, France
example, offers VMs in two main markets: on-demand and spot. On-demand VMs can be deployed at any time, offering high availability since they cannot be interrupted by Amazon provider while allocated by a user. On the other hand, spot VMs are unused EC2 resources with a huge discount (according to Amazon the discount can be up to 90% when compared to on-demand prices) but can be revoked and terminated by Amazon whenever it requires the resources back. Since December 2017, Amazon EC2 has defined a VM allocation policy where spot prices are more stable and with little differences over the days, i.e., they do not vary according to users’ resource requests demand [23]. Furthermore, Amazon EC2 has introduced the spot VM hibernation feature that hibernates a spot VM instead of terminat
Data Loading...
