摘要 |
PROBLEM TO BE SOLVED: To provide a device and a method for deciding a clustering coefficient for a database by using block level sampling. SOLUTION: These device and method perform block level sampling of the database, process the data of the database to generate one or a plurality of lines, and process one or a plurality of lines to generate a clustering coefficient of a selected index. Further, if a device and a method of a desirable performance form are used, a distribution of clustering coefficients over a certain range can be decided, and it is thereby possible to discriminate a range where clustering coefficients are high and a range where clustering coefficients are low. Search memory paging performance using the existing or potential index corresponding to the sampled data can be predicted by using the clustering coefficient distribution, and then, database search performance using the existing or potential index can be predicted with respect to particular database inquiry.
|