In‑Depth Analysis of Redis GEOADD and GEORADIUS Commands and Their Algorithmic Complexity

Redis provides geospatial indexing using six commands—GEOADD, GEOPOS, GEODIST, GEOHASH, GEORADIUS, and GEORADIUSBYMEMBER—allowing developers to build high‑performance “nearby people” features.

GEOADD adds a member (latitude, longitude, name) to a sorted set by converting the coordinates into a 52‑bit geohash score and invoking ZADD. The source code in geo.c validates arguments, extracts coordinates, encodes them with geohashEncodeWGS84 , creates score objects, and finally calls zaddCommand to store the member.

GEORADIUS (and its variant GEORADIUSBYMEMBER) queries members within a given radius around a point. The command parses optional flags such as WITHDIST, WITHCOORD, WITHHASH, ASC/DESC, COUNT, STORE, and STOREREDIST, then computes the search area by converting the radius to a geohash bounding box and selecting the nine surrounding geohash cells.

The core functions geohashGetAreasByRadiusWGS84 , membersOfAllNeighbors , and geoGetPointsInRange locate candidate members by performing range scans on the underlying sorted‑set skip‑list (or ziplist), followed by precise spherical distance checks using geohashGetDistanceIfInRadiusWGS84 .

Using a nine‑cell grid avoids boundary edge cases and dramatically reduces the number of elements examined, because each cell’s geohash range is contiguous and can be scanned efficiently. The overall time complexity of a radius query is O(N + log M), where N is the number of results returned and M is the number of elements examined in the nine cells, yielding very high performance in memory‑resident Redis.