Paper Title: Improving the locality of generalized integrated interleaved codes

Paper Abstract: To enable the continued scaling of distributed storage, it is essential to adopt locally recoverable (LRC) erasure codes that allow failures to be recovered with much shorter latency and smaller penalty on the network bandwidth. Compared to other LRC codes, the generalized integrated interleaved (GII) codes achieve better tradeoffs on locality, redundancy, and complexity. This talk presents two methods for further improving the locality of GII codes. The first utilizes a lower-zero-triangular nesting matrix to construct two-layer GII codes. Fewer symbols are accessed to recover from fewer failures. The second generalizes three-layer integrated interleaved codes that have better locality than two-layer codes. The parities shared by the codewords are used in a more flexible way to recover from more failures.

Paper Author: Xinmiao Zhang, Associate Professor, Ohio State University (ECE Department)

Author Bio: Xinmiao Zhang received her Ph.D. degree from the University of Minnesota in 2005. She joined The Ohio State University as an Associate Professor in 2017. Prior to that, she was a Senior Technologist at Western Digital. Her research spans the areas of digital storage and communications, VLSI architecture design, security, and cryptography. She is currently the Vice-Chair from Academia of the Data Storage Technical Committee of the IEEE Communications Society, and was the Vice-Chair from Industry for the 2017-2018 term. She also serves as a member of the Board of Governors of the IEEE Circuits and Systems Society, as well as IEEE CASCOM, VSA, and DISPS technical committees. She has served on the committees of many conferences, including ISCAS, SiPS, ICC, NVMW, ICASSP, GLOBECOM, and GLSVLSI. She is a recipient of the NSF CAREER Award in 2009.