This paper studies how to minimize the total cost of answering r queries over n elements in an online manner (i.e., the next query is given only after the previous query’s result is ready) when the value r ≤ n is unknown in advance. Traditional indexing, which first builds a complete index on the n elements before answering queries, may be unsuitable because the index’s construction time - usually Ω(n log n) - can become the performance bottleneck. In contrast, for many problems, a lower bound of Ω(n log (1+r)) holds on the total cost of r queries for every r ∈ [1, n]. Matching this lower bound is a primary objective of deferred data structuring (DDS), also known as database cracking in the system community. For a wide class of problems, we present generic reductions to convert traditional indexes into DDS algorithms that match the lower bound for a long range of r. For a decomposable problem, if a data structure can be built in O(n log n) time and has Q(n) query search time, our reduction yields an algorithm that runs in O(n log (1+r)) time for all r ≤ (n log n)/(Q(n)), where the upper bound (n log n)/(Q(n)) is asymptotically the best possible under mild constraints. In particular, if Q(n) = O(log n), then the O(n log (1+r))-time guarantee extends to all r ≤ n, with which we optimally settle a large variety of DDS problems. Our results can be generalized to a class of "spectrum indexable problems", which subsumes the class of decomposable problems.
@InProceedings{tao:LIPIcs.ICDT.2025.10, author = {Tao, Yufei}, title = {{Maximizing the Optimality Streak of Deferred Data Structuring (a.k.a. Database Cracking)}}, booktitle = {28th International Conference on Database Theory (ICDT 2025)}, pages = {10:1--10:18}, series = {Leibniz International Proceedings in Informatics (LIPIcs)}, ISBN = {978-3-95977-364-5}, ISSN = {1868-8969}, year = {2025}, volume = {328}, editor = {Roy, Sudeepa and Kara, Ahmet}, publisher = {Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik}, address = {Dagstuhl, Germany}, URL = {https://drops.dagstuhl.de/entities/document/10.4230/LIPIcs.ICDT.2025.10}, URN = {urn:nbn:de:0030-drops-229512}, doi = {10.4230/LIPIcs.ICDT.2025.10}, annote = {Keywords: Deferred Data Structuring, Database Cracking, Data Structures} }
Feedback for Dagstuhl Publishing