Fig. 1
From: Comparison of LSM indexing techniques for storing spatial data
LSM flush operation. Primary (double line) and secondary (single line) memory components are flushed independently to the top of disk components of the corresponding index. Delete table is flushed together with the corresponding memory component (\(P_M\) and \(P_D\), \(S_M\) and \(S_D\)), creating antimatter (a.k.a tombstone) records (marked with -) in the flushed component. Index keys are underlined. The primary index P’s schema is (CarID, OwnerID, ManufactureYear). A secondary index S is built on OwnerID