Big Data is a new class of technology that gives businesses more insight into their massive data sets, allowing them to make better business decisions and satisfy customers. Big data systems are also a desirable target for hackers due to the aggregation of their data. Hadoop is used to handle large data sets through reading and writing application programs on a distributed system. Hadoop Distributed File System is used to store massive data. Since HDFS does not safeguard data privacy, encrypting the file is the right way to protect the stored data in HDFS but takes a long time. In this paper, regarding privacy concerns, we use different compression-type data storage file formats with the proposed user-defined function (XOR-Onetime pad with AES) to secure data in HDFS. In this way, we provide a dual level of security by masking the selective data and whole data in the file. Our experiment demonstrates that the whole process time is significantly smaller than that of a conventional method. The proposed UDF with ORC, Zlib file format gives 9-10% better performance results than 2DES and other methods.  Finally, we decreased the load time of secure data and significantly improved query processing time with the Hive engine.