From NoSQL to Mo' SQL - Adding Structure and Queriability to Riak
summary
Riak Time Series adds native SQL by collocating time-sliced rows via quantized time buckets and a two-part key so queries touch a few nodes instead of the whole ring. A DDL-like create table emits dynamic Erlang modules that teach the gossip-based cluster about schema and key logic. Queries compile to distributed subqueries run by a distributed coordinator, trading even load for queryability. Future work explores CRDT-backed schemas and delta CRDTs to make large sets queryable and reduce I/O.
AI-written summary · about
chapters (15)
- 0:00Intro
- 4:36Data Access Nodes
- 6:28Time Series
- 8:54Tradeoffs
- 13:10Sequel
- 14:39Riak Architecture
- 16:17Structured Data
- 18:41Query System
- 21:27Decorative Properties
- 23:41Decorative Sequel
- 25:57Fragments
- 26:57Execution Hints
- 28:08The Future
- 28:41CRT Sets
- 34:12Pushing Work to the Edge
