Create flexible indexes

Quickly and efficiently search lookups on high-dimension vector data with StarTree. Ideal for use cases around GenerativeAI (GenAI), Retrieval Augmented Generation (RAG), chatbots and Natural Language Processing (NLP), and recommendation systems.

Build an intelligent materialized view for pre-computing certain aggregations for a wide range of dimension filters while improving query latency and query throughput. Star-Tree Index, unique to Apache Pinot, provides an innovative way to control the degree of materialization, thus allowing the user to trade off query latency versus storage overhead.

Handle structured and semi-structured data while enabling extremely fast query processing on highly nested JSON columns. Users can analyze free structured JSON or text documents and ingest nested JSON columns as-is — without needing to preprocess or transform them in any way.

Perform powerful text search queries on arbitrary text data by configuring a Text Index to accelerate query processing.

Applied to a pre-sorted column, enabling efficient range queries and binary search, reducing the need for scanning unnecessary data blocks.

Stores min-max values for column segments, allowing fast pruning of irrelevant data during range queries.

Maps unique column values to the row IDs where they appear, allowing for fast filtering and lookups, especially on high-cardinality columns.

In Pinot, you can use the geospatial index to optimize lookups based on geospatial data. This includes geospatial data types, such as point, line, and polygon; geospatial functions for querying spatial properties and relationships; and geospatial indexing, used to process spatial operations.

A Bloom filter is a space-efficient probabilistic data structure that is used to test whether an element is a member of a set. IN Pinot, Bloom filter helps prune segments that do not contain any record matching an equality predicate.

Pinot provides a timestamp index type for optimizing queries based on time series data. This data type stores value as a millisecond-long epoch value internally. Typically for analytics queries, you don’t need this low level of granularity, as scanning the data and time value conversion can result in costly implications given the big size of data. The time series index optimizes these queries.

A lightweight index that selectively indexes only a subset of data, reducing storage overhead while still enabling faster lookups. It is particularly useful for optimizing queries on large datasets where full indexing would be too costly.