Optimized Design And Implementation Of Cloud Storage Platform For Mass Fingerprint Data In Fingerprint Payment System

In recent years, with the rapid development of ecommerce the transaction volume online is increasing, which puts forward higher requirements on the security of Internet transactions. Some scholars want to use the fingerprint identification technology to improve the security of online transactions. Compared to the traditional alphanumeric passwords,fingerprint recognition is more convenient and not easy to be cracked.Compared to the traditional password way, fingerprint recognition brings us not only high security but also high complexity. The system needs to store more data, and needs to improve the efficiency of read and write about data.Then the traditional relational database has been unable to meet our needs, so this paper uses the cloud storage platform to deposit mass fingerprint data in the fingerprint payment system.This paper chooses HDFS as the infrastructure cloud storage platform, but there are still some problems when we use it to storage mass fingerprint data in the fingerprint payment system.On the one hand, HDFS is used to process a large number of large files at the beginning of the design,so there will be some problems when we use it to storage mass fingerprint data.For example,the NameNode’s out of memory and the low efficiency query.On the other hand,the load balancing algorithm in the HDFS only one-sided to consider the disk space,so it did not consider the data node’s state comprehensive and can not make the clusters operation more healthy.In view of the above two problems, this paper focuses on the optimization scheme when we use HDFS to storage mass fingerprint data and multiple angles load balancing algorithm.so the main work as follows:(1)Analysis the demand of cloud storage platform in the fingerprint payment system,and proposes the goal of design according to the analysis about demand.(2)Design the client of upper management. First of all, design the modules, and then use the HTML language to program the Web client.(3)We optimized the HDFS based on the characteristic of data which need to be stored.The data need to be transmitted to the small file discriminant combiner before upload it to the HDFS.Small files are merged according to geographical information.The metadata is distributed storaged in the DataNode.Set the caching policies for hot files.(4)Optimize the load balancing algorithm in HDFS,proposed the multi angle load balancing algorithm which is dynamic.We not only need to considers the utilization of DataNode’s disk memory,but also need to consider how often the file to be read and the different allocation of different DataNode when we judge the load state of a DataNode.Integrated several dynamic angles,we can get the threshold.We should balance the load of DataNode at the right time. We choose not to balance the load when the cluster is overweight in order to avoid increasing of cluster load.The work we have done above indeed improved the performance of cloud storage platform which has been proved by experiments.The work also make the cluster’s load balancing more comprehensive.The research work we have done in this paper have actively significance for the construction of cloud storage platform in the fingerprint payment system.