RediSearch 2.8 release notes

Requirements

RediSearch v2.8.4 requires:

• Minimum Redis compatibility version (database): 7.2
• Minimum Redis Enterprise Software version (cluster): 7.2.4

v2.8 GA (v2.8.4) (July 2023)

This is the General Availability release of RediSearch 2.8.

Headlines

RediSearch 2.8 introduces support for RESP3, new features, performance improvements, and bug fixes.

What’s new in 2.8.4

This new major version introduces new and frequently asked for Geo Polygon Search, adding the GEOSHAPE field type that supports polygon shapes using WKT notation. Besides the current GEO (alias for GEOPOINT) used in geo range queries, we add support for POLYGON and POINT as new geo shape formats (new GEOSHAPE). In addition, 2.8 brings performance improvements for SORT BY operations using FT.SEARCH and FT.AGGREGATE, and new FORMAT for enhanced responses on FT.SEARCH and FT.AGGREGATE in RESP3 only.

Features:

• Introduce support for geo polygon shapes and queries:

  • Adding GEOSHAPE field type to map polygons in the SCHEMA on FT.CREATE (MOD-4798)

  • Support for polygons POLYGON and POINT using WKT notation, for example POLYGON((lon1 lat1, lon2 lat2, ...))

  • Adjust the query syntax on FT.SEARCH for polygons using the predicate @geom:[OPERATOR $poly] and defining polygon in WKT format as PARAMS 2 poly "POLYGON((10 20, ...))" using DIALECT 3

  • Initially WITHIN and CONTAINS operators with GEOSHAPES for now

  • Support multiple coordinate systems: cartesian (X,Y) with the flag FLAT for flat earth and geographic (lon, lat) using the flag SPHERICAL (MOD-5303). Geographic coordinate system using spherical indexing as default (SPHERICAL)

  • Add memory usage per Geometry Index in the FT.INFO response report (MOD-5278)

• Introduce performance optimization for sorting operations on FT.SEARCH and FT.AGGREGATE as default on DIALECT 4. It will improve performance in 4 different scenarios, listed below:

  • Skip Sorter - applied when there is no sort of any kind. The query can return once it reaches the LIMIT requested results.

  • Partial Range - applied when there is a SORTBY a numeric field, with no filter or filter by the same numeric field, the query iterates on a range large enough to satisfy the LIMIT requested results.

  • Hybrid - applied when there is a SORTBY a numeric field in addition to another non-numeric filter. Some results will get filtered, and the initial range may not be large enough. The iterator then is rewinded with the following ranges, and an additional iteration takes place to collect LIMIT requested results.

  • No optimization - If there is a sort by score or by non-numeric field, there is no other option but to retrieve all results and compare their values.

• Add WITHCOUNT argument that allow return accurate counts for the query results with sorting. This operation processes all results in order to get accurate count, being less performant than the optimized option (default behavior on DIALECT 4) (MOD-5311)

• New FORMAT argument in FT.SEARCH and FT.AGGREGATE to retrieve the results as JSON strings or RESP3 hierarchical structures (RESP3 only) (MOD-5390)

Improvements (since 2.8.3):

• #3717 - Polygon shapes validation and orientation correction when clockwise (MOD-5575)

• #3534 - Vector Similarity [0.7.0]

• #3657 - Allow GC calls for all tiered indexes in the schema

• #3701 - HNSW is now using data blocks to store vectors and metadata instead of array

Changed behavior:

• #3355, #3635 Expired keys deleted from replica’s index, returning an empty array instead of nil (MOD-4739)