Dual-Code Bounds on Multiple Concurrent (Local) Data Recovery

TL;DR

This paper explores bounds on the ability of linear redundancy storage schemes to support concurrent local data recovery, using generator matrix properties to establish limits on data access rates.

Contribution

It introduces dual distance outer bounds for such systems, providing sharp bounds for certain classes of matrix families within classical coding theory.

Findings

Derived bounding polytope for data access rates

Established two dual distance outer bounds

Bounds are sharp for specific matrix classes

Abstract

We are concerned with linear redundancy storage schemes regarding their ability to provide concurrent (local) recovery of multiple data objects. This paper initiates a study of such systems within the classical coding theory. We show how we can use the structural properties of the generator matrix defining the scheme to obtain a bounding polytope for the set of data access rates the system can support. We derive two dual distance outer bounds, which are sharp for some large classes of matrix families.