分析型数据库动态数据脱敏技术应用

Practice of Dynamic Data Masking Techniques in Analytical Databases

在数字化转型过程中,需要兼顾数据的价值利用和用户的隐私安全。近年来,国内外陆续出台了相关法律,要求对个人信息通过技术手段进行保护。数据脱敏技术是保障隐私数据安全性及促进数据价值共享的关键技术,在对用户数据匿名化、去标识化的过程中,保留一定的原始数据分布特征和格式属性,使得应用无法通过脱敏后的数据反向推断原始数据,但是仍然可以利用数据的分析价值。本文介绍了中兴通讯EBASE-A分析型数据库中通过改写查询语法树实现数据脱敏的方法,讨论了其技术路线优势、实现原理及架构设计。实验显示,应用典型脱敏规则会产生2.5%~15.7%的额外性能消耗,在应用中符合预期。

In the process of digital transformation,it is necessary to balance the value utilization of data and the privacy of users.In recent years,relevant laws and regulations have been introduced,requiring the pro-tection of personal information through technical means.Data masking technology is a key technique that en-sures the security of privacy data while facilitating the sharing of data value.In the process of anonymizing and de-identifying user data,it retains certain original data distribution characteristics and format attributes,mak-ing it impossible for applications to reversely infer the original data from the masked data,while still leveraging the analytical value of the data.This paper introduces the method of data masking in ZTE EBASE-A analytical database by rewriting the query tree,and discusses its technical advantages,implementation principles,and architecture design.Experimental results show that typical masking rules cause additional performance over-head ranging from 2.5%to 15.7%,within an acceptable range for applications.