Index Suggestion based on Log Analysis Widya Wilhelmina Simbolon (a*), Sonya Ahnela Pardosi (b), Elisabeth Jesica Gurning (b)
a) Information System, Del Institute of Technology
Sitoluama, Lagu Boti, Kabupaten Toba Samosir, Sumatera Utara 22381, Indonesia
*widyawilhelmina[at]gmail.com
b) Information System, Del Institute of Technology
Sitoluama, Lagu Boti, Kabupaten Toba Samosir, Sumatera Utara 22381, Indonesia
Abstract
SELECT operations are the most frequent operations in the database. This operation performs a scan on the table to find the record corresponding to the executed query. Optimization of queries with indexing can enhance the record search process by forming a data structure to find one or a group of records efficiently without having to scan all records in the table. The analysis of log file queries on the database is one of the ways to define the records that need to be indexed. This is because a log query records all the queries executed in the database. In this research, the query log file need to be parsed to find statistical information access to the attributes in the database. Then, the statistical information will be used in forecasting process to estimate the growth of access on that attribute. The forecasting method used in this research is Support Vector Regression (SVR). The SVR method with the RBF (Radial Basis Function) kernel can provide indexing suggestions based on the statistical information of each attribute obtained from the query log file. From the results of the index suggestion development, the parsing process takes a long time to produce the necessary statistical information.
