Computational Storage for Data Integrity and Security (MSST ‘24)

Research project undertaken by myself and a Computer Engineering Ph.D student for an advanced data storage systems class. The goal of this project is to research and analyze the viability of using computational storage devices for accelerating tasks common to data integrity and ransomware protection schemes. We have identified research gaps in erasure coding and data security tasks and their applicability to computational storage devices and developed a proof-of-concept library (CSDGuard) using C++ and OpenCL that implemented I/O operations, computations, and performance measurement utilities common to those tasks. We then synthesized and optimized a hardware-based accelerator kernel using AMD Vitis High-Level Synthesis tools, and were able to observe up to a 70% performance increase utilizing CSDGuard over a reference software-based implementation. Our work was accepted to be a part of the poster presentation portion of the 2024 Massive Storage Systems and Technology conference in Santa Clara, CA.