SPIKE: Best practice generation for storage area networks

Abstract

This paper presents SPIKE, an automated algorithm to generate best practices by analyzing Storage Area Network (SAN) configuration errors. Best practices are a useful tool in problem diagnosis as most configuration problems are caused by the violation of best practices in the storage network domain. However, the manual generation of best practices is tedious, error-prone and costly in terms of time and manpower. SPIKE uses a combination of information-retrieval principles, entity ranking and decision-tree classification to statistically infer the best practices for the prevention of SAN configuration problems. Preliminary results from an initial implementation of SPIKE indicate speed and accuracy improvements over manually generating best practices.