Using Data Entanglement to Propagate Redundancy

When: Thursday, July 14, 2016
Where: PGH 563
Time: 11:00 AM

Speaker: Veronica Estrada Galinanes, PhD Candidate at University of Neuchatel, NE, Switzerland; Visiting Scholar, Storage Systems Research Center, UC Santa Cruz, CA

Host: Dr. J.F. Paris

Imagine if a librarian could have recovered all lost works of Aristotle by shuffling together the pages of other books written by his predecessors and successors. Unless the librarian applies some magical trick or alchemist knowledge, shuffling pages will not bring any new content.

In general, current storage systems can tolerate two or three failures by storing data redundantly. The risk of missing some content forever is high when all replicas become unreachable. Digital information opens more opportunities than classical replication. Data pieces can be mixed and combined together in multiple ways to recreate different contents.

My research explores a different way of achieving redundancy by constructing a large texture of redundant data. We propose data entanglement to increase system reliability. Additionally, the method can improve system performance. In this talk, I will present the benefits and limitations of this new method.

Bio:

Ms. Estrada joined the Complex Systems Group at the University of Neuchatel, Switzerland in 2012, where she works on her PhD under the supervision of Professor Pascal Felber. Currently, she is visiting the Storage Systems Research Center, UC Santa Cruz, where she is working on erasure coding algorithms in collaboration with Dr. Ethan Miller. Her interests include systems reliability, data entanglement, security and systems architecture. She holds a Master degree in Applied Computer Science from University of Tokyo and an M.S. in Electronic Engineering from University of Buenos Aires. Before she started building a scientific career, she worked several years in the industry, particularly in the oil and gas industry and for Argentina's federal government.