Catching Waves with Time-Series Data
summary
Specific claims and numbers: a homegrown system Little Table stores time-ordered metrics and clustered disk layout and Cisco moroi reports 320 terabytes on it; benchmarks claim Time scale DB loads 1 billion rows in 1/15th the time of vanilla Postgres and yields ~396× faster queries on a 100M-row test. Concrete engineering choices include composite keys ordered by prefix, append-only writes, compression vs aggregation, and a retention plan (base table monthly TTL, daily/weekly/monthly aggregates with 6m/1y/5y TTLs) and API selection rules.
AI-written summary · about
chapters (15)
- 0:00Introduction
- 1:07Time series data importance
- 3:10Choosing time series tools
- 8:27Building time series solution
- 10:33Efficient time series organization
- 13:20Composite keys for performance
- 15:50Optimal time series queries
- 17:57Data retention with compression
- 20:03Row vs column databases
- 21:50Aggregation strategies
- 23:58API endpoints design
- 26:26Visualizing time series data
- 27:27Advanced tools integrations
- 28:52Real-world application lessons
- 29:33Closing remarks
related talks
29mCatching Waves with Time-Series Data
51mPostgreSQL, The Time-Series Database You Want
49mPostgreSQL, The Time-Series Database You Actually Want
51mBuilding a scalable time-series database on PostgreSQL
39mTimescaleDB: Re-engineering PostgreSQL as a time-series database
58mIntroduction to Timeseries Databases
39mA New Data Model for Time Series Data Processing- 39mA New Data Model for Time Series Data Processing