作者: Sangho Yi , Derrick Kondo
DOI: 10.1007/978-3-642-17913-6_11
关键词: Spare part 、 Resource (project management) 、 Distributed computing 、 Supply and demand 、 Bidding 、 Cloud computing 、 Bid price 、 Middleware 、 Provisioning 、 Computer science
摘要: With the recent advent of Cloud Computing, vision computing as a utility has reached new heights. Resource prices can now change dynamically based on current demand and supply. In December 2009, Amazon released Spot Instances, which sell spare capacity their data centers. This is sold using an auction-like bidding system. Instances are allocated or terminated depending whether instance price above below user’s bid price. While pricing reliability vary dramatically dynamically, middleware have few mechanisms that deal with cost-reliability trade-offs effectively. Also failure rates highly number bids, degree bid, available resources. this paper, we present how checkpointing be used to reduce cost volatility resource provisioning. Based real, history spot instances show performance schemes including proposed one in terms monetary completion time tasks.