Research On Virtual Computing Environment And Its Parallel Scheduling Strategy

Virtual computing environment can organize a large number of idle computing resourceson the network, and shield the differences in the structure and performance of computingresources, by organizing resources to work together and complete the calculation of thelarge-scale data processing tasks, to provide users with a transparent and unified resourceaccess interface. This paper designs and implements a seismic-data-processing orientedvirtual computing environment, and based on the computing features of pre-stack depthmigration, we proposed a seismic data processing oriented parallel scheduling algorithm.The seismic-data-processing oriented virtual computing environment, applying thecoordinated structure as the organization of resources, and combining the Web architectureand JAVA framework technology to implements the basic architecture of the virtualcomputing environment, which has a simple structure and is flexible and easy to extend, so asto reach the goal of low-cost and high-efficiency virtual computing environment. The paralleltask scheduler, in line with the characteristics of the seismic data processing, is the core of theparallel partitioning and scheduling of large-scale tasks, which can be continually used tostudy and applied to actual large-scale seismic data processing.The parallel scheduling algorithm for seismic data processing, based on virtualcomputing environment and combined with the parallel characteristics of the seismic dataprocessing, tasks can be divided according to the characteristics of the set of availableresources and assigned to the appropriate resource. The computational complexity of tasksassigned to a particular node, is proportional to the computing power of the node. So thealgorithm considering the computing power as the standard of the division of tasks, with thecontrol of load balancing to minimize the impact caused by the performance difference ofresources, and also the overall parallel execution time.