Achieving Market Liquidity Through Autonomic Cloud Market Management

Abstract

Due to the large variety in computing resources, Cloudmarkets often suffer from a low probability of findingmatches between consumers' bids and providers' asks,resulting in low market liquidity. The approach of servicelevel agreement (SLA) templates (i.e., templates forelectronic contracts) is a mean to reduce this variety asit channels the demand and supply. However, until now, theSLA templates used were static, not able to reflect changesin users' requirements. To address this shortcoming, weintroduce an adaptive approach for automatically derivingpublic SLA templates based on the requirements of marketparticipants. To achieve this goal, we utilize clusteringalgorithms for grouping similar requirements and learningmethods for adapting the public SLA templates to observedchanges of market conditions. To assess the benefits of theapproach, we conduct a simulationbased evaluation andformalize a utility and cost model. Our results show thatthe use of clustering algorithms and learning algorithmsimproves the performance of the adaptive SLA templateapproach.