One of the most important issues of database storage management techniques is indexing. A comparison of several indexing structures for complex object in object-oriented database systems was studied, based on assumption of non-uniform distribution of attribute values.

While conventional index structures deal with one-dimensional data have well-known properties, the multi-dimensional stuctutures represent an arena of intensive research worldwide. One of the operations in spatial access based systems are spatial joins. The effective algorithm for spatial joins which combines both I/O and CPU cost optimization was developed.

A modification of ARIES/NT algorithm for nested transactions rollback and recovery was developed. This modification allows to perform the forward recovery for advanced transaction models. The algorithm was implementated at the storage system level. The transaction control protocols for real-time database systems were also studied.

Several indexing techniques for advanced database systems targeted to new application areas were investigated. In particular, a new indexing structure and algorithm for full text systems was proposed and evaluated. The proposed algorithm provides for efficient updates of inverted lists. Dynamic updates are especially important for applications in the Internet.

An important feature for improving query evaluation performance in object-oriented databases is high-level indexing support. An evaluation of update performance of access support relations was performed, resulting in new performance estimations. Previous results were related only to retrieval performance of this indexing structure.

Indexing structures for temporal databases were proposed and investigated.