Meeting service level objectives of Pig programs.

ABSTRACTCloud computing offers a compelling platform to access a large amount of computing and storage resources on demand. As the technology matures, service providers have started shifting their focus to support additional user requirements such as QoS guarantees and tailored resource provisioning for achieving service performance goals. An increasing number of MapReduce applications associated with live business intelligence require completion time guarantees. We aim to solve the resource provisioning problem: given a Pig program with a completion time goal, estimate the amount of resources (a number of map and reduce slots) required for completing the program with a given (soft) deadline. We develop a simple yet elegant performance model that provides completion time estimates of a Pig program as a function of allocated resources. Then this model is used as a basis for solving the inverse resource provisioning problem for Pig programs. We evaluate our approach using a 66-node Hadoop cluster and a popular PigMix benchmark. The designed performance model accurately estimates the required amount of resources for Pig programs with completion time goals: the completion times of the Pig programs with allocated resources are within 10% of the targeted deadlines.