Google Cloud Spanner: Global consistency at scale
summary
Cloud Spanner is a fully managed, horizontally scalable SQL database that delivers global serializable consistency by using TrueTime GPS+atomic clocks to bound clock uncertainty (epsilon), minimizing waits for commits. Instances allocate 2 TB per node and a regional 3‑node instance is replicated across three zones (effectively nine replicas). Features include online schema changes, stale reads up to 15s to serve local replicas, and write semantics where a row write is per‑cell (indexes add cells) with a 20,000 mutations/transaction limit.
AI-written summary · about
chapters (28)
- 0:00Introduction
- 0:53Agenda
- 1:23Backstory
- 2:17What did we need
- 4:44Open Standards
- 5:31Cloud Spanner vs other databases
- 9:14Cloud Spanner use cases
- 11:33Tables
- 12:26Regional Instance
- 13:06Global Consistency
- 15:32Owners
- 16:55Throughtime
- 18:40Hardware
- 20:24Query
- 21:50stale consistent reads
- 24:04readwrite transection
- 25:39data model
- 26:29interleaving
- 28:00SQL interleaving
- 28:26Schema donts
- 29:57Example
- 30:30Sharded counters
- 30:44Chart counters
- 31:01Getting started
- 32:20Creating a database
- 34:45Example schema
- 37:41Demo
- 43:52Cap Theorem
related talks
33mGoogle Cloud Spanner: Global consistency at scale
37mGoogle Cloud Spanner: Going global with relational database
41mSpanner - Google's Distributed Database
49mSpanner - fully managed horizontally scalable relational DB with ACID transactions
22mOrder And Consistency At Scale
01h 08mKeynote: Building Spanner
24mnative databases and storage for games
01h 02mDiamond Sponsor Hall