FASTEN: An FPGA-Based Secure System for Big Data Processing

Security is a major concern in adopting cloud computing especially for processing private and/or sensitive data. FASTEN is an FPGA-based secure system for processing big data. It leverages the security features in modern FPGAs such as crypto engines and PUF. In FASTEN, the kernel functions for data processing are translated into hardware using a High Level Synthesis and are executed inside the FPGA fabric. Thus, the plaintext data and security keys are not exposed to main memory and secondary storage, keeping security-critical data from malicious insiders and outsiders. We have constructed a 24-node cluster using Zynq-7000 FPGA-SoCs, and conducted the performance evaluation with K-means, LWLR, and Sobel filter applications using Hadoop MapReduce on Linux. Experiment outcomes and security analysis reveal that the FASTEN provides both performance and security advantages over the system with Hadoop’s native security support, at the expense of additional hardware.