Practice and Exploration on Privacy-Preserving Computing on Enterprise-level Big Data

Big data is nowadays at the core of businesses, and it is one of the most valuable assets for every organization (company). Since big data is generally sensitive to businesses and privacy, performing big data analytics (e.g., revealing hidden patterns or identifying secret correlations) efficiently without jeopardizing the data’s security and privacy is crucial to enterprises. JD.com is building a privacy-preserving computing platform/framework for processing large-scale data securely by combining Intel® Software Guard Extensions (SGX) with several other security technologies. We jointly developed the remote attestation clustering services with the BigDL team. With the adaptation of BigDL PPML(Privacy-Preserving Machine Learning), our platform provides a Trusted Cluster Environment for standard Big Data & AI applications. It is capable of handling terabytes of big data while preserving data confidentiality and privacy, setting it apart from existing privacy-enhanced computation techniques that lack similar characteristics for possible real-world business applications. We are making this solution a potential candidate to ensure end-to-end security for the entire distributed workflow in both zero-trust intranets, and untrusted public clouds.