LCS is an efficient data eviction strategy for Spark, a popular distributed in-memory data processing system. LCS gets dependencies information between cache data via analyzing application, and calculates the recovery cost during running - View it on GitHub