A Learning Technique for VM Allocation to Resolve Geospatial Queries

Provisioning of virtual machines for efficient geospatial query management on cloud is an interesting and challenging work. The aim of this paper is to distribute workloads of different types of spatial queries into suitable virtual machine efficiently. To increase the effectiveness of the system serving geospatial queries, we use real-time geospatial query pattern learning methodology. This methodology is used to train the application specific properties and the system will learn which type of the geospatial query should be allocated what type of virtual machine automatically. The learning methodology gives knowledge about the resource required by each type of geospatial query. Using this understanding, various geospatial query templates are stored in the query template repository for further assistance. This way fast and robust assignment of virtual machine for the geospatial queries is possible which reduces their waiting time.