Research On Task Scheduling Algorithm For Generalized Forestry Geographic Information Processing In Cloud Computing

With the rapid development of Internet and the Internet of Things technology, the temporal granularity of forestry geospatial information refines gradually,the scale of data grows rapidly and types of data becomes more and more complex and diverse over time, the extension of geographic information gives birth to generalized forestry geographic information. Generalized forestry geographic information includes not only survey and mapping geographic information, but also includes the forest ecosystem information collected by sensor network and the network text which contains forestry geospatial information etc. Massive generalized forestry geographic information is a huge to traditional GIS, and also brings a demand for generalized forestry GIS, GIS relying on a single node to process information model has become more and more difficult to deal with data processing tasks in time, while a efficient multi-node distributed cloud computing technology processed large-scale data is expected to solve the problem, therefore is important to study on generalized forestry geographical information by cloud computing technology.The key for sloving the requirement of multi-user’s large number of tasks being executed efficiently in cloud computing environment is task scheduling algorithms. Good scheduling strategy can often balance the load, load balancing scheduling strategy can ensure that all data processing resource be dispatched fairly,avoid the platform bottlenecks caused by load imbalance, improves customer satisfaction, improve node resources’utilization and overall system performance.Considering the question of generalized forestry geographic information is executed efficiently in cloud computing environment, the work as follows in this paper.(1) Research on cloud computing technology, analyzes the Hadoop cloud computing platform, proposes the concept of generalized forestry geographic information and its technologies demands, analyzes the forest region tourists’track data on Hadoop platform by MapReduce programming, the experimental result shows that compares with stand-alone environment, hadoop cluster environment is more efficient on processing capacity, it also means that it is feasible to efficient processing generalized forestry information.(2) The key for sloving large number of tasks being executed efficiently in cloud computing environment is task scheduling algorithms, task scheduling algorithm under a cloud environment is studied in this paper, analyses the existing priority-based scheduling algorithms and load balancing algorithms, and proposes an improved dynamic priority-based load balancing algorithm TS-PFB. The dynamic priority of this algorithm depends on value density and urgency of the implementation, the priority increases with time to ensure tasks can be delivered in time to different users. By simulating the behavior of fireflies, calculates the decision variable considered attractiveness (the earliest completion time) and fluorescence brightness (load restraint) to balance the load on VM. To verify the performance of the algorithm, the compare experiments about all task completion time, the success rate and load balancing degree in CloudSim simulation environment had been completed, and the results show that the algorithm has a better scheduling performance.(3) Finally, this paper designs and implements dynamic priority-based scheduling policy in hadoop, the scheduling results show that the scheduling strategy can complete more Jobs in dadline.