Research On Monitoring Platform For Agricultural Products Quality And Resource Scheduling Algorithm Based On Hadoop

In recent years, the problem that the quality and safety of agricultural products arises endlessly has caused widespread concern of society. Therefore it is urgent to construct a system to monitor the quality of agricultural products. The solution of supervising that problem is to construct the village information monitoring platform. It is not feasible in practice for traditional techniques to use a single node database backend to collect and process the data of the quality of agricultural products for that the velocity of processing and responding data can be very slow when facing the big data, so it’s difficult to break through the current bottleneck of agricultural product quality supervision encountered in the process. Thus, this paper focuses on the issue of applying related technology of cloud computing on the platform to supervise the safety of agricultural products and put forward the overall framework with resource scheduling strategy to build this platform.To construct the overall framework of the platform, this paper describes the function and its related technology methodologies of each layer, makes a detailed design for storage and processing of data distributed in this platform and designs the function of monitoring the excessive property of agricultural products data. Then, this paper applies the parallel processing and distributed storage system of Hadoop in the platform for processing and storing data, studies and compares the advantages and disadvantages of FIFO scheduling, fair scheduling and capacity scheduling systematically, puts forward a RDA scheduling strategy based on the three scheduling strategy above and implements the RDA scheduling strategy. Compared with the three scheduling strategy of Hadoop, the RDA scheduling strategy improves the judgment of the behind task thus making the speculative execution more efficiently, adjusts the size of the job queue automatically under the multi-job and multi-task environment thus adjusting the load level of the clusters dynamically, divides the nodes of the cluster into two types thus making task execution more efficient. At the end, the simulation results show that when facing the multi-job and multi-task environment, the RDA scheduling strategy has a advantage over other three scheduling strategies and it has stronger applicability and more advance for the platform of monitoring safety of agricultural products.This paper puts forward the construction scheme of applying the cloud computing technology in the platform of monitoring safety of agricultural products thus making monitoring feedback of agricultural products more precise and timely makes the monitoring more efficient by designing the Map-Reduce function for monitoring and early warning of data of agricultural products. In addition, this paper improves the three scheduling strategies in Hadoop, designs the dynamic adjusting algorithm at running time for monitoring of the platform, thus, making the platform keep staying on a higher resource utilization. The research of this paper provides a certain level of technical support for building the early warning systems of safety of agricultural products.