| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 2.86 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
With Cloud Computing, access to computational resources has become increasingly facilitated and applications could offer improved scalability and availability. The datacenters that support this model have a huge energy consumption and a limited pricing model. One way of improving energy efficiency is by reducing the idle time of resources - resources are active but serve a limited useful business purpose. This can be done by improving the scheduling across datacenters. We present FairCloud, a scalable Cloud-Auction system that facilitates the allocation by allowing the adaptation of VM requests (through conversion to other VM types and/or resource capping - degradation), depending on the User profile. Additionally, this system implements an internal reputation system, to detect providers with low Quality of Service (QoS). FairCloud was implemented using CloudSim and the extensions CloudAuctions. FairCloud was tested with the Google Cluster Data. We observed that we achieved more quality in the requests while maintaining the CPU Utilization. Our reputation mechanism proved to be effective by lowering the Order on the Providers with lower quality.
Description
Keywords
Cloud computing Energy Pricing Auctions Scheduling Reputation User profiles
Citation
FONSECA, Artur; SIMÃO, José; VEIGA, Luís – FairCloud: truthful cloud scheduling with continuous and combinatorial auctions. In OTM Confederated International Conferences On the Move to Meaningful Internet Systems. Valletta, Malta: Springer, 2017. ISBN 978-3-319-69461-0. Vol. 10574, pp. 68-85
Publisher
Springer Verlag