Thursday, November 12th
8:35-10:05
Session A-9: Computational Storage Increases System Throughput and Scalability (Computational Storage Track)
Organizer: JB Baker, Senior Director of Product Management, ScaleFlux

Paper Title: Annual Update on Computational Storage

Paper Abstract: As data stores grow larger, the simple act of moving them around absorbs way too much compute time and bus capacity. The solution is to move compute closer to where the data resides. A popular way of doing this is to simply attach a programmable processor to each storage unit, an approach called computational storage. Several methods have emerged involving FPGAs, standard processors, and other computing units. They offer a wide range of cost/performance tradeoffs and varied implementations for deployment, provisioning, management, and security. Meanwhile, the SNIA Computational Storage Technical Work Group has been promoting both standards and interoperability. Likely future developments include more software support in leading operating systems, more packaged software or reference designs to handle common tasks, and simpler ways to attach compute to standard SSDs. The result should be far higher cost-performance ratios and far more scalability to solve big data problems in real-time analysis, AI/ML, high-performance computing, and signal and image processing.

Paper Author: Chuck Sobey, Chief Scientist, ChannelScience

Author Bio: Through over 20 years of providing clarity on the complexities of emerging data storage technologies, Chuck Sobey has earned a reputation as a trusted technology advisor with key engineers, scientists, and executives from practically every storage-related company. He has advised and contributed to important developments in hard disk drives, tape drives, magneto-optics, laser optics, Blu-ray Disc, biomedical technology, 3D NAND flash, spin-transfer torque magnetic RAM, and more. He presented the keynote address at the industry’s celebration of the 50th anniversary of the hard disk drive; gave the opening technical talk at China’s first flash forum in Beijing; and wrote the insightful foreword to the recent textbook 3D Flash Memories. Chuck is a long-time member of the Flash Memory Summit’s executive committee, for which he has developed very popular tracks covering controller technologies, ECC, and the growing mainland China flash industry. Chuck has electrical and computer engineering degrees from Carnegie Mellon University and the University of California at Santa Barbara. He also presents free seminars (“Hirable for Life”) for university students and experienced engineers and scientists on thriving in the ever-changing career of engineering.