ποΈ Indexes
Indexes can help quickly locate data that matches certain query conditions. To be specific, when an index is built on a table based on some of the table's columns, queries that use these columns as query conditions require scanning only on portions of the table data in order to locate the data that meet the conditions. As such, query efficiency is improved. StarRocks provides a variety of index types, mainly divided into the following two categories:
ποΈ Prefix indexes
Specify one or more columns to comprise the sort key at table creation. The data rows in the table will be sorted based on the sort key and then stored on the disk.
ποΈ Bitmap indexes
This topic describes how to create and manage a bitmap index, along with usage cases.
ποΈ Bloom filter indexes
This topic describes how to create and modify bloom filter indexes, along with how they works.
ποΈ N-gram bloom filter index
The N-gram bloom filter index is a special Bloom filter index which is typically used to accelerate the LIKE queries and the calculation speed of the ngramsearch and ngramsearchcase_insensitive functions.
ποΈ Full-text inverted index
Since version 3.3.0, StarRocks supports full-text inverted indexes, which can break the text into smaller words, and create an index entry for each word that can show the mapping relationship between the word and its corresponding row number in the data file. For full-text searches, StarRocks queries the inverted index based on the search keywords, quickly locating the data rows that match the keywords.
ποΈ Vector Index
This topic introduces the vector index feature of StarRocks and how to perform an approximate nearest neighbor search (ANNS) with it.