Veronica (Vero) is currently studying computer science as a graduate student at the University of Neuchâtel. Her PhD thesis is about data encoding for reliable storage systems and proposes entanglement codes. Vero spent six months of her doctoral studies doing research at the Storage Systems Research Center, University of California, Santa Cruz. She holds a master’s degree from the University of Tokyo and an electronic engineer diploma from the University of Buenos Aires. In Tokyo, she studied anomalous behavior observed in network traffic by classifying data using artificial neural networks and intrusion detection systems. Before leaving for Japan, she worked for many years in the industry and public administration sector in Buenos Aires, Argentina.
In the future, Vero plans to continue working in research and development for innovative, high-tech solutions. During her free time, she enjoys dancing tango, art, photography and cooking.
Practical Codes for Storage Systems »
Project: Failure Repair Dynamics Study
Goal of the project:
Understand the behavior of entanglement codes using simple computational models like cellular automata.
Summary of work:
Entanglement codes are a technique to create redundant information in the system. They tangle new data blocks with old ones, building entangled data chains that are woven into a growing mesh of interdependent content.
This project is about the development of a research framework to study failure repair dynamics on storage systems. The program allows playing with code settings and distinct failure assumptions.
The figures shown above illustrate the dynamics of a storage system under constant failure injection. Each cell represents a storage unit. The color black indicates that the cell is available. The grid is subdivided in columns four cells wide, indicated with pink lines. The cell at the left of the line represents data, while the other cells represent redundant information. At each step, a certain percentage of failures are injected to the available cells, coloring the cell with “white,” while the system keeps trying to recover the missing cells.
Results and future work:
This technique seems a good way to study entanglement codes as well as other codes used in storage systems. I plan to continue this research direction to compare entanglement with other codes. In addition, I would like to improve the failure model and include repair constraints.