Title: Distributed storage systems from combinatorial designs
Speaker: Aditya Ramamoorthy, ECpE Associate Professor
Abstract: Large scale distributed storage systems are used in several contexts, such as social networking sites (Facebook, Google+ etc.) and cloud storage (Amazon, Microsoft etc.). The widespread adoption of these systems has introduced several new research problems. For instance, these systems often consist of individually unreliable nodes; however the system needs to have reliable data access. When nodes fail, the system needs to be repaired in a speedy manner, by consuming as few resources (drives accessed, energy etc.) as possible. In recent years there has been a lot of interest in the design of regenerating codes which are erasure codes targeted towards distributed storage systems. In this talk, I will provide a broad overview of this area.
Following this, I will discuss our recent work on the regenerating codes from combinatorial designs. Combinatorial design theory has its roots in recreational mathematics and is concerned with the arrangement of elements of a finite set into subsets, such that the collection of subsets has certain “nice” properties. For instance, the Kirkman schoolgirls problem posed in 1850 asks: “Fifteen young ladies in a school walk out three abreast for seven days in succession: it is required to arrange them daily so that no two shall walk twice abreast”. It turns out that the solutions to these seemingly puzzle like problems, can be used to arrive at distributed storage systems with several desirable features. I will discuss some of our recent work that presents constructions from Steiner systems, affine geometries, Hadamard designs, mutually orthogonal Latin squares and high-girth graphs.