Hybrid spot instance based resource provisioning strategy in dynamic cloud environment

Abstract

Abstract-Utilization of resources to the maximum extent in large scale distributed cloud environment is a major challenge due to the nature of cloud. Spot Instances in the Amazon Elastic Compute Cloud (EC2) are provisioned based on highest bid with no guarantee of task completion but incurs the overhead of longer task execution time and price. The paper demonstrates the last partial hour and cost overhead that can be avoided by the proposed strategy of Hybrid Spot Instance. It aims to provide reliable service to the ongoing task so as to complete the execution without abruptly interrupting the long running tasks by redefining the bid price. The strategy also considers that on demand resource services can be acquired when spot price crosses on-demand price and thereby availing high reliability. This will overcome the overhead involved during checkpointing, restarting and workload migration as in the existing system, leading to efficient resources usage for both the providers and users. Service providers revenue is carefully optimized by eliminating the free issue of last partial hour which is a taxing factor for the provider.Simulation carried out based on real time price of various instances considering heterogenous applications shows that the number of out-of-bid scenarios can be reduced largely which leads to the increased number of task completion. Checkpointing is also minimized maximally due to which the overhead associated with it is reduced. This resource provisioning strategy aims to provide preference to existing customers and the task which are nearing the execution completion.

Spot pricing in the Cloud ecosystem: A comparative investigation

“…However, the existing fault-tolerance techniques are not free [SMG14]. For example, checkpointing would inevitably occupy extra fixed-price resource (e.g., storage), while duplication and migration would require more compute resources.…”

“…There is also a special concern about Amazon's policy of not charging customers for the interrupted partial hours, if the spot service usage is terminated by Amazon [3]. In the worst case, the uncharged time rises up to 30% of a spot instance's total run time [DTM12], which would become a burden and reduce the provider's profit [SMG14]. However, since this is Amazon's own policy, we do not treat this concern as a limitation of spot pricing in a generic sense.…”

“…To relieve the overhead burden of not charging the last partial hour, the study [SMG14] proposed a Hybrid Spot Instance approach that could share the fault tolerance cost between the service provider and consumers, and meanwhile could make the spot service more reliable. The Hybrid Spot Instance approach is essentially based on the aforementioned checkpointing technique, and it allows stretching the user bid till checkpointing is done when an out-of-bid situation occurs.…”

Price forecasting for spot instances in Cloud computing

“…In the literature, many algorithms have been designed to save the rental cost of resources with fixed prices for bag of tasks [22,23], MapReduce tasks [24], workflows [25,26,27] and bag-of-task based workflows [28]. [31] allowing cloud users to modify bids after an initial bid has been set. Sadashiv et al [31] assumed that bids can be adjusted at any time for avoiding out-of-bid events.…”

Cloud capacity planning and HSI based optimal resource provisioning

“…The aim of HSI based resource provisioning is to improve the reliability of current SI users by prohibiting abrupt termination during out-of-bid situation. This is performed by enabling users to stretch their bid price till checkpointing is carried out or to perform rebid process until the user's bid price equals the on-demand price [13]. Such a bidding approach is enforced in ebay as a measure to avoid rebidding.…”