Creating spatial autocorrelation via clustering
When spatial data is loaded into a table, data is usually organized depending on the order of the incoming rows. This ordering of the data on disk has a direct impact on the performance of the spatial queries. Consider a simple SDO_ANYINTERACT
query that retrieves all the data inside a rectangular box. After the spatial query is performed, the rows that satisfy the result are retrieved from the table. If all of these rows are spread over different data blocks, the cost of the query increases, as many blocks have to be fetched to form the result set. If data in the blocks can be organized in such a way to minimize the number of blocks fetched for each query, the query performance would improve. This improvement will be greater for queries that fetch a large number of rows for each query. For example, in web mapping applications, small scale maps show less detail and large scale maps show more detail. As the scale goes from small to large, more...