Time-series data, or data being associated with its respective time of occurrence, is everywhere. From the obvious cases, such as metrics, observability, IoT data, all the way to logs, invoicing, or payment records. While storing some of these in relational databases is standard practice, people often reach for specific time-series databases when volume gets high. But imagine if you could have all of them in the same database: PostgreSQL.
13. Time-Related Observations
of a
Subject
CREATE TABLE measurements (
time TIMESTAMPTZ NOT NULL,
deviceID INT NULL,
sensor_value REAL NULL,
location geometry NULL
);
CREATE TABLE stocks (
time TIMESTAMPTZ NOT NULL,
price INT NULL,
symbol TEXT NULL
);
14. NYSE transactions per day NYSE transactions per second NYSE transactions in 5 years
4.2 B 162 K 5.4 Trillion
illion
Consolidated Tape A
trading volume
https://www.nyse.com/markets/us-equity-volumes
4.2B Transactions
23400 Trading Seconds
CFTC requires that
fi
nancial records be kept
for 5 years and “readily
available” for 2 years
18. • Established Ecosystem
• Rock-Solid
• Production Proven
• SQL (no custom … something language)
• Multi-Model Support (through Extensions)
• JOINing between all your Data
19. • Established Ecosystem
• Rock-Solid
• Production Proven
• SQL (no custom … something language)
• Multi-Model Support (through Extensions)
• JOINing between all your Data
• I Love Elephants!
29. Blog Posts
What Is Time-Series Data?
Hacking NFL Data with PostgreSQL, TimescaleDB, and SQL
What Is Time-Series Forecasting?
https://tsdb.co/blog-time-series-data
https://tsdb.co/blog-n
fl
-data
https://tsdb.co/blog-what-is-time-series-forecasting
Docs https://tsdb.co/timescale-docs