Research Article
Optimization of Query Processing Through Nested Persistent Storage Implementation of Relational Tables
Issue:
Volume 10, Issue 3, September 2026
Pages:
82-100
Received:
8 June 2026
Accepted:
22 June 2026
Published:
28 July 2026
DOI:
10.11648/j.ajist.20261003.11
Downloads:
Views:
Abstract: Relational databases use physical design techniques such as indexing, clustering, and partitioning to improve performance without altering the logical schema seen by applications. This paper proposes nesting of persistent storage structures as a new physical optimisation technique. Related tables can be implemented as a single nested persistent storage structure, eliminating join operations when processing complex queries and removing redundant foreign key repetitions on the many side of the relationship. The paper describes two independent nesting strategies. The schema driven approach selects nesting candidates from the logical schema before any workload is known. This selection subject to storage optimisation and database organisation constraints, formulating the selection as a constrained maximum-weight branching problem. The workload driven approach determines nesting decisions subject to aggregate query processing frequencies, applying nesting transformations in descending order of join frequency. To preserve logical transparency, mapping metadata records each attribute's physical access path. The paper shows how SQL queries expressed over the logical schema can be automatically transformed into equivalent queries over the nested physical structures, keeping the nested representation hidden from applications. A prototype implementation in PostgreSQL on the TPC-H benchmark demonstrates the feasibility of the proposed approach and confirms that standard indexing techniques can recover the query performance overhead introduced by nesting. By reducing storage redundancy and eliminating join operations while remaining compatible with existing relational interfaces, the technique preserves the relational programming model.
Abstract: Relational databases use physical design techniques such as indexing, clustering, and partitioning to improve performance without altering the logical schema seen by applications. This paper proposes nesting of persistent storage structures as a new physical optimisation technique. Related tables can be implemented as a single nested persistent sto...
Show More
,
Janusz Getta
