Regenerating codes are a class of codes proposed for efficient repair of failed nodes in distributed storage systems. In this paper, we address the fundamental problem of handling errors and erasures which may occur during data reconstruction and node repair in regenerating codes. There are numerous scenarios which motivate this problem such as time-critical data recovery, dynamic load balancing, and security from malicious adversaries. We provide outer bounds, and explicit regenerating codes achieving these bounds for a wide range of system parameters. This also establishes the capacity of these systems for these parameter regimes.
View on arXiv