A Two-tier Coarse Indexing Scheme for MLS Database Systems

In this paper, we propose a two-tier indexing scheme for multilevel secure database systems, primarily with the intent of improving query response time and reducing the storage required for indexing. At the bottom tier, our scheme requires separate single-level indices, one for each partition of the multilevel relation; at the top tier, our scheme requires a coarse multilevel index consisting of only those key values from the single-level indices that are necessary to direct a query to the appropriate single-level index. Our scheme seems suitable for both single-level and range queries. We prove this claim for B+trees by providing a detailed performance analysis for this index structure. We also give the algorithms for inserting and deleting key values, as well as for searching for a key value, in the proposed index structure.