Fast and Flexible JSON Retrieval in Cassandra Using SAI
summary
Method stores the JSON blob and creates a “super‑shredded” row per document that indexes paths and typed values so CQL + SAI can answer filters like existence, numeric ranges and text on nested fields. Queries push multiple terms into one WHERE so SAI does parallel joins; maps and sets are indexable. Updates that need max/add-to-set require a select, modify and conditional write (read‑modify‑write) and inserts use lightweight transactions. Measured overhead: 3.5 ms extra for HTTP/JSON roundtrip and 1.4 ms coordinator read on a 3M‑row SAI index. SAI indexes on maps and sets and super-shredding enable the API but write cost and missing operators (eg not‑contains) remain.
AI-written summary · about
chapters (11)
- 0:00JSON retrieval in Cassandra
- 2:23JSON data model
- 6:00JSON filtering expressions
- 10:16JSON updates and RMW
- 11:20Secondary indexes and LWT
- 15:13Super shredding for JSON
- 17:37Super shredding implementation
- 20:47Updating JSON docs
- 23:40SII and transaction improvements
- 24:42Performance benchmarks
- 26:28Demo and next steps
