Computing and Information Systems - Theses
Permanent URI for this collection
Search Results
1 - 2 of 2
-
ItemResource provisioning in spot market-based cloud computing environments( 2014)Recently, cloud computing providers have started offering unused computational resources in the form of dynamically priced virtual machines (VMs), also known as "spot instances". In spite of the apparent economical advantage, an intermittent nature is inherent to these biddable resources, which may cause VM unavailability. When an out-of-bid situation occurs, i.e. the current spot price goes above the user's maximum bid, spot instances are terminated by the provider without prior notice. This thesis presents a study on employing cloud computing spot instances as a means of executing computational jobs on cloud computing resources. We start by proposing a resource management and job scheduling policy, named SpotRMS, which addresses the problem of running deadline-constrained compute-intensive jobs on a pool of low-cost spot instances, while also exploiting variations in price and performance to run applications in a fast and economical way. This policy relies on job runtime estimations to decide what are the best types of spot instances to run each job and when jobs should run. It is able to minimise monetary spending and make sure jobs finish within their deadlines. We also propose an improvement for SpotRMS, that addresses the problem of running compute-intensive jobs on a pool of intermittent virtual machines, while also aiming to run applications in a fast and economical way. To mitigate potential unavailability periods, a multifaceted fault-aware resource provisioning policy is proposed. Our solution employs price and runtime estimation mechanisms, as well as three fault tolerance techniques, namely checkpointing, task duplication and migration. As a further improvement, we equip SpotRMS with prediction-assisted resource provisioning and bidding strategies. Our results demonstrate that both costs savings and strict adherence to deadlines can be achieved when properly combining and tuning the policy mechanisms. Especially, the fault tolerance mechanism that employs migration of VM state provides superior results in virtually all metrics. Finally, we employ a statistical model of spot price dynamics to artificially generate price patterns of varying volatility. We then analyse how SpotRMS performs in environments with highly variable price levels and more frequent changes. Fault tolerance is shown to be even more crucial in such scenarios.
-
ItemMeta scheduling for market-oriented grid and utility computing( 2010)Grid computing enables the sharing and aggregation of autonomous IT resources to deliver them as computing utilities to end users. The management of the Grid environment is a complex task as resources are geographically distributed, heterogeneous and autonomous in nature, and their users are self-interested. In utility-oriented Grids, users define their application requirements and compete to access the most efficient and cheapest resources. Traditional resource management systems and algorithms are based on system-centric approaches which do not take into account individual requirements and interests. To this end, market-oriented scheduling is an adequate way to solve the problem. But current market-oriented systems generally, either try to maximise one user’s utility or one provider’s utility. Such approaches fail to solve the problem of contention for cheap and efficient resources which may lead to unnecessary delays in job execution and underutilisation of resources. To address these problems, this thesis proposes a market-oriented meta-scheduler called “Meta-Broker”, which not only coordinates the resource demand but also allocates the best resources to users in terms of monetary and performance costs. The thesis results demonstrate that considerable cost reduction and throughput can be gained by adopting our proposed approach. The meta-broker has a semi-decentralised architecture, where only scheduling decisions are made by the meta-broker while job submission, execution and monitoring are delegated to user and provider middleware. This thesis also investigates market-oriented meta-scheduling algorithms which aim to maximise the utility of participants. The market-oriented algorithms consider Quality of Service (QoS) requirements of multiple users to map jobs against autonomous and heterogeneous resources. This thesis also presents a novel Grid Market Exchange architecture which provides the flexibility to users in choosing their own negotiation protocol for resource trading. The key research findings and contributions of this thesis are: - The consideration of QoS requirements of all users is necessary for maximising users’ utility and utilisation of resources. The uncoordinated scheduling of applications by personalised user-brokers leads to overloading of cheap and efficient resources. - It is important to exploit the heterogeneity between different resource sites/data centers while scheduling jobs to maximise the provider’s utility. This consideration not only reduce energy cost of computing infrastructure by 33% on average, but also enhance the efficiency of resources in terms of carbon emissions. - By considering both system metrics and market parameters, we can enable more effective scheduling which maximises the utility of both users and resource providers.