Hongyi Wang, Bingsheng He, Vijayan Prabhakaran, and Lidong Zhou
29 June 2009
Storage systems such as file systems and database systems maintain certain structures to facilitate efficient access to the data. Corruptions of those structures could be particularly damaging, leading to data loss far exceeding the number of bits that are actually corrupted.
Repairing structural corruptions is an important line of defense against data loss. Unlike previous corruption repair mechanisms (such as fsck), which are often ad hoc and heuristic-based, we propose a new approach named Crystal that revolves around the notion of a structure. With the notions of structural Hamming distance and structural edit distance, Crystal defines the problem of corruption repair as a global optimization problem, allows the incorporation of old system snapshots for repair, and advocates the inclusion of structural robustness to corruptions in the design of the system structure, rather than handling corruptions as an after-thought.
In Proceedings of the Fifth Workshop on Hot Topics in System Dependability (HotDep'09)