Rack-Aware Regenerating Codes With Multiple Erasure Tolerance

We study the rack-aware storage system where all nodes are organized in racks and within each rack can communicate freely without taxing bandwidth. Rack-aware regenerating codes (RRCs) were proposed for minimizing repair bandwidth single erasures. In initial setting of RRCs, a node requires participation remaining containing failed as well large number helper no failures. Consequently, may be infeasible front multiple this work, relaxed model that tolerate failures by simultaneously reducing intra-rack connections cross-rack is proposed. A tradeoff between under derived, parameters two extreme points on curve characterized minimum respectively. Moreover, corresponding to explicitly constructed over fields size comparable code length with lowest sub-packetization. Finally, convenience practical use, systematic encoding processes also established.