Erasure coding is a low-cost redundancy mechanism for distributed storage systems by
storing stripes of data and parity chunks. Wide stripes are recently proposed to suppress the …

We propose repair pipelining, a technique that speeds up the repair performance in general
erasure-coded storage. By carefully scheduling the repair of failed data in small-size units …

Erasure coding is widely used for massive storage in data centers to achieve high fault
tolerance and low storage redundancy. Since the cross-rack communication cost is often …

We study a generalization of the setting of regenerating codes, motivated by applications to
storage systems consisting of clusters of storage nodes. There are n clusters in total, with m …

The paper is devoted to the problem of erasure coding in distributed storage. We consider a
model of storage that assumes that nodes are organized into equally sized groups, called …

A new system model reflecting the clustered structure of distributed storage is suggested to
investigate interplay between storage overhead and repair bandwidth as storage node …

Erasure codes offer a storage-efficient redundancy mechanism for maintaining data
availability guarantees in storage clusters, yet also incur high network traffic consumption …

Erasure coding is a storage-efficient redundancy scheme for modern clustered storage
systems by storing stripes of data and parity blocks across the nodes of multiple clusters; in …

Erasure coding becomes a practical redundancy technique for distributed storage systems
to achieve fault tolerance with low storage overhead. Given its popularity, research studies …

As a common choice for fault tolerance in today's storage systems, erasure coding is still
hampered by the induced substantial traffic in repair. A variety of erasure codes and repair …