The new release of GeoCouch improves spatial index building time by up to 10x, thanks to a new bulk insertion algorithm that uses an append-only data structure and reduces the number of disk operations required for inserting geometry entries. The current implementation updates the index when a request is made to use it, rather than when a new document is inserted, allowing for more efficient update performance. GeoCouch's bulk insertion algorithm creates clusters out of the entire bulk update that match the existing index structure and then bulk loads them into R-trees using an overlap minimizing top-down (OMT) algorithm. The algorithm balances the height of the target tree and reduces physical writes to disk, resulting in improved performance and file sizes compared to traditional single item insertion strategies. Future improvements may involve optimizing the client API for setBulk method or having Couchbase Server execute bulk operations.