TL;DR: AI gaming company Latitude replaced Amazon Aurora with TimescaleDB for their interactive storytelling platform, cutting storage costs by $12K monthly while improving query performance.
--> Jump to full story
What is TimescaleDB? An open-source database built on PostgreSQL, optimized for time-series data and real-time analytics while maintaining full SQL compatibility.
The Real-Time Data Challenge ๐
Latitude's AI Dungeon platform generates dynamic storylines for thousands of players simultaneously. As their user base grew, Amazon Aurora couldn't efficiently handle their time-series data needs:
- Rising storage costs
- Degrading query performance
- Challenges with real-time analytics
- Engineering resources consumed by database management
The Solution?๐ก
Latitude migrated from Amazon Aurora to Timescale.
"TimescaleDB by itself is what allows for the hypertables, the query performance, and the compression. That's where it's something like $12,000 monthly storage savings."
Key TimescaleDB Features:
- Hypertables for efficient time-series partitioning
- Continuous aggregates for real-time analytics
- Native compression to reduce storage requirements and costs
-- Example of hypertable creation in TimescaleDB
CREATE TABLE events(
time TIMESTAMPTZ NOT NULL,
player_id INTEGER,
action_type TEXT,
context JSONB
);
-- Convert to hypertable
SELECT create_hypertable('events', 'time');
Performance Results
- $12,000 monthly database storage cost reduction
- Improved query performance despite growing data volumes
- Seamless real-time analytics powering their AI storytelling engine
- Capacity to handle larger datasets as player base expands
For Latitude's engineering team, TimescaleDB delivered the rare combination of better performance and lower costs, allowing them to focus on creating immersive AI gaming experiences instead of database optimization.
Discussion ๐ฌ
Have you faced similar scaling challenges with time-series data? What database solutions have worked for your real-time analytics needs?
If you're already using Postgres, and/or you think Timescale (Postgres optimized for time series) may be a the best database solution for your workload, try our open-source TimescaleDB extension
Install TimescaleDB
Install from a Docker container:
- Run the TimescaleDB container:
docker run -d --name timescaledb -p 5432:5432 -e POSTGRES_PASSWORD=password timescale/timescaledb:latest-pg17
- Connect to a database:
docker exec -it timescaledb psql -d "postgres://postgres:password@localhost/postgres"
See other installation options or try Timescale Cloud for free.
Top comments (0)