Explicit Construction of Minimum Bandwidth Rack-Aware Regenerating Codes
In large data centers, storage nodes are organized in racks, and the cross-rack communication dominates the system bandwidth. We explicitly construct codes for exact repair of single node failures that achieve the optimal tradeoff between the storage redundancy and cross-rack repair bandwidth at the minimum bandwidth point (i.e., the cross-rack bandwidth equals the storage size per node). Moreover, we explore the node repair when only a few number of helper racks are connected. Thus we provide explicit constructions of codes for rack-aware storage with the minimum cross-rack repair bandwidth, lowest possible redundancy, and small repair degree (i.e., the number of helper racks connected for repair).
READ FULL TEXT