Concurrency control for adaptive indexing

Abstract

Adaptive indexing initializes and optimizes indexes incrementally, as a side effect of query processing. The goal is to achieve the benefits of indexes while hiding or minimizing the costs of index creation. However, index-optimizing side effects seem to turn readonly queries into update transactions that might, for example, create lock contention.This paper studies concurrency control in the context of adaptive indexing. We show that the design and implementation of adaptive indexing rigorously separates index structures from index contents; this relaxes the constraints and requirements during adaptive indexing compared to those of traditional index updates. Our design adapts to the fact that an adaptive index is refined continuously, and exploits any concurrency opportunities in a dynamic way.A detailed experimental analysis demonstrates that (a) adaptive indexing maintains its adaptive properties even when running concurrent queries, (b) adaptive indexing can exploit the opportunity for parallelism due to concurrent queries, (c) the number of concurrency conflicts and any concurrency administration overheads follow an adaptive behavior, decreasing as the workload evolves and adapting to the workload needs.

An experimental evaluation and analysis of database cracking

“…This research group has proposed several different indexing techniques to address different dimensions of database cracking, including updates [14], tuple reconstruction [15], convergence [16], concurrency control [8,9], and robustness [11]. In this paper, we critically review database cracking in several aspects.…”

“…Additionally, we inspect a recently published work [23], which identifies CPU efficiency problems in the standard cracking algorithm and proposes alternatives. Furthermore, we investigate the current state-of-the-art in parallel cracking algorithms [2,8,9,23] and compare them against each other. Finally, we also extend the experimental parameters previously used in database cracking, e.g.…”

“…Parallelizing cracking algorithms As exploiting modern hardware implies using the multi-threading capabilities of the system, we investigate in this section how the cracking algorithms can be parallelized. To do so, we first re-evaluate a lock-based parallel standard cracking algorithm [8,9] that serializes the crack-in-two operation at the granularity of partitions of data for inter-query parallelism. Additionally, we add the algorithms we proposed in our study on parallel adaptive indexing algorithms [2] and put the methods under a new set-up to the test.…”

“…Thus, several researchers have proposed adaptive indexing techniques to cope with these requirements. In particular, Database Cracking has emerged as an attractive approach for adaptive indexing in recent years [8,11,[14][15][16][17][18]. Since the release of our conference paper [25] on which this work builds upon, three more studies have been published [2,9,23].…”

Scaling up analytical queries with column-stores

“…Orthogonally to these families of techniques, the data reorganization performed by column-stores during query execution (for example by database cracking [15]) generates both reads and writes, but with an appropriate concurrency control mechanism [12] it shows the desired behavior. Concurrency in the optimization phase.…”