Parsing Set of Queries to Obtain Parse Matrix for Table Index Evaluator and Recommender
Shefali Trushit Naik · SSRN Electronic Journal · 2018
The use of appropriate indexing improves the performance of transactions in heterogeneous distributed database, whereas inappropriate or no indexing deteriorates the same. Properly designed index leads to faster data access, which ultimately improves the execution of transactions. Various Relational Database Management Systems (RDBMS) and third-party tools exist, which provide suggestion for index management, but up to certain limits. These tools provide index suggestion with limited and simple queries. They do not analyze or suggest index for aggregate queries, sub-queries and other complicated queries. The applications which access data from heterogeneous databases need an index evaluator and recommender to analyze and recommend indexes for tables. For this type of multiple heterogeneous distributed databases, Table Index Evaluator and Recommender (TIER) is proposed which takes set of queries as inputs. Queries in the set are based on local and remote tables (tables which are distributed on various RDBMS). In order to recommend indexes, the fields which are mentioned in WHERE and HAVING clauses of inputted set of queries should be parsed. Besides this, the total frequency of each field table-wise and overall is required. The parsed fields with frequency result in Parse Matrix (PM) and the obtained PM is used by TIER for further processing. In this paper, the algorithm to obtain Clause Matrix (CM) and PM is described.