timescaledb

Overview

Version

Build

You can build the RPM / DEB packages for timescaledb using pig build:

pig build pkg timescaledb         # build RPM / DEB packages

Install

You can install timescaledb directly. First, make sure the PGDG and PIGSTY repositories are added and enabled:

pig repo add pgsql -u          # Add repo and update cache

Install the extension using pig or apt/yum/dnf:

• Install
• pig
• dnf
• apt

pig install timescaledb;          # Install for current active PG version

pig ext install -y timescaledb -v 18  # PG 18
pig ext install -y timescaledb -v 17  # PG 17
pig ext install -y timescaledb -v 16  # PG 16
pig ext install -y timescaledb -v 15  # PG 15

dnf install -y timescaledb-tsl_18       # PG 18
dnf install -y timescaledb-tsl_17       # PG 17
dnf install -y timescaledb-tsl_16       # PG 16
dnf install -y timescaledb-tsl_15       # PG 15

apt install -y postgresql-18-timescaledb-tsl   # PG 18
apt install -y postgresql-17-timescaledb-tsl   # PG 17
apt install -y postgresql-16-timescaledb-tsl   # PG 16
apt install -y postgresql-15-timescaledb-tsl   # PG 15

Preload:

shared_preload_libraries = 'timescaledb';

Create Extension:

CREATE EXTENSION timescaledb;

Usage

Sources: README, TimescaleDB 2.27.0 release, CREATE TABLE API, create_hypertable() API, continuous aggregate API, add_columnstore_policy() API, GUCs

timescaledb is a PostgreSQL extension for time-series and event analytics. The current docs emphasize CREATE TABLE ... WITH (tsdb.hypertable), continuous aggregates, automation jobs, and moving chunks into the columnstore.

Hypertables

CREATE EXTENSION timescaledb;

CREATE TABLE ts_test (
  ts timestamptz NOT NULL,
  id bigint,
  v integer
) WITH (
  tsdb.hypertable,
  tsdb.orderby = 'ts DESC'
);

To convert an existing PostgreSQL table, use the generalized hypertable API:

CREATE TABLE ts_existing (
  ts timestamptz NOT NULL,
  id bigint,
  v integer
);
SELECT create_hypertable('ts_existing', by_range('ts'));

• CREATE TABLE ... WITH (tsdb.hypertable) has been documented since TimescaleDB 2.20.0 and is the best-practice path for new hypertables.
• For TimescaleDB 2.23.0 and later, the first TIMESTAMP or TIMESTAMPTZ column is selected automatically as the partition column unless more than one candidate makes the choice ambiguous.
• create_hypertable() still works for converting existing tables.

Continuous aggregates and jobs

CREATE MATERIALIZED VIEW ts_hourly
WITH (timescaledb.continuous) AS
SELECT time_bucket('1 hour', ts) AS bucket,
       count(*) AS cnt,
       avg(v)   AS avg_v
FROM ts_test
GROUP BY bucket;

SELECT add_continuous_aggregate_policy(
  'ts_hourly',
  start_offset => INTERVAL '3 hours',
  end_offset => INTERVAL '1 hour',
  schedule_interval => INTERVAL '1 hour'
);

SELECT add_job('user_defined_action', '1h');

• Continuous aggregates require time_bucket(...) on the hypertable’s time dimension.
• The continuous aggregate WITH clause supports timescaledb.materialized_only; the current API default is TRUE, so real-time aggregation is not enabled unless configured otherwise.

Columnstore

CREATE TABLE crypto_ticks (
  "time" timestamptz,
  symbol text,
  price double precision,
  day_volume numeric
) WITH (
  tsdb.hypertable,
  tsdb.segmentby = 'symbol',
  tsdb.orderby = 'time DESC'
);

CALL add_columnstore_policy('crypto_ticks', after => INTERVAL '60 days');

• CREATE TABLE ... WITH (tsdb.hypertable) enables columnstore by default unless tsdb.columnstore = false.
• add_columnstore_policy() replaces the older add_compression_policy() API and requires either after or created_before, not both.
• Bloom filters are enabled by default for new columnstore chunks. Existing chunks need recompression before they have bloom indexes.

Relevant GUCs

SET timescaledb.enable_direct_compress_insert = on;
SET timescaledb.enable_cagg_rewrites = on;
SET timescaledb.enable_columnar_scan_filter_pushdown = on;

timescaledb.enable_direct_compress_insert and timescaledb.enable_direct_compress_copy enable tech-preview direct compression during ingestion. TimescaleDB 2.27.0 adds timescaledb.enable_cagg_rewrites and timescaledb.cagg_rewrites_debug_info, and documents timescaledb.enable_columnar_scan_filter_pushdown as enabled by default.

Caveats

• This project’s CSV tracks TimescaleDB 2.27.0 for PostgreSQL 15-18.
• TimescaleDB 2.27.0 adds Hypercore columnstore performance work: vectorized filters, bloom-filter pruning for UPDATE/DELETE equality predicates, and bloom-filter pruning for UPSERT.
• The 2.27.0 release notes list backward-incompatible upgrade caveats for affected compressed int2 bloom sparse indexes and for composite bloom metadata generated by v2.26.
• The 2.27.0 release notes announce that the June 2026 TimescaleDB release is planned to be the last version supporting PostgreSQL 15.