Background

The nature of time-series data is that the relevance of information diminishes over time. If stale data is no longer required, users can delete old data from QuestDB to either save disk space or adhere to a data retention policy. This is achieved in QuestDB by removing data partitions from a table.

This page provides a high-level overview of partitioning with examples to drop data by date. For more details on partitioning, see the partitioning page.

Strategy for data retention

A simple approach to removing stale data is to drop data that has been partitioned by time. A table must have a designated timestamp assigned and a partitioning strategy specified during a CREATE TABLE operation to achieve this.

:::note

Users cannot alter the partitioning strategy after a table is created!

:::

Tables can be partitioned by one of the following:

• DAY
• MONTH
• YEAR

CREATE TABLE my_table(ts TIMESTAMP, symb SYMBOL, price DOUBLE) timestamp(ts)
PARTITION BY DAY;

Dropping partitions

:::caution

Use DROP PARTITION with care as QuestDB cannot recover data from dropped partitions!

:::

To drop partitions, users can use the ALTER TABLE DROP PARTITION syntax. Partitions may be dropped by:

• DROP PARTITION LIST - specifying a comma-separated list of partitions to drop

  --Delete a partition
  ALTER TABLE my_table DROP PARTITION LIST '2021-01-01';
  --Delete a list of two partitions
  ALTER TABLE my_table DROP PARTITION LIST '2021-01-01', '2021-01-02';

• WHERE timestamp = - exact date matching by timestamp

  ALTER TABLE my_table DROP PARTITION
  WHERE timestamp = to_timestamp('2021-01-01', 'yyyy-MM-dd');

• WHERE timestamp < - using comparison operators (< / >) to delete by time range relative to a timestamp. Note that the now() function may be used to automate dropping of partitions relative to the current time, i.e.:

  --Drop partitions older than 30 days
  WHERE timestamp < dateadd('d', -30, now())

Usage notes:

• The most chronologically recent partition cannot be deleted
• Arbitrary partitions may be dropped, which means they may not be the oldest chronologically. Depending on the types of queries users are performing on a dataset, it may not be desirable to have gaps caused by dropped partitions.

Example

The following example demonstrates how to create a table with partitioning and to drop partitions based on time. This example produces 5 days’ worth of data with one incrementing LONG value inserted per hour.

CREATE TABLE my_table (timestamp TIMESTAMP, x LONG) timestamp(timestamp)
PARTITION BY DAY;
INSERT INTO my_table
SELECT timestamp_sequence(
    to_timestamp('2021-01-01T00:00:00', 'yyyy-MM-ddTHH:mm:ss'),100000L * 36000), x
FROM long_sequence(120);

For reference, the following functions are used to generate the example data:

• timestamp sequence with 1 hour stepping
• row generator with long_sequence() function which creates a x:long column

The result of partitioning is visible when listing as directories on disk:

my_table
├── 2021-01-01
├── 2021-01-02
├── 2021-01-03
├── 2021-01-04
└── 2021-01-05

Partitions can be dropped using the following query:

--Delete days before 2021-01-03
ALTER TABLE my_table DROP PARTITION
WHERE timestamp < to_timestamp('2021-01-03', 'yyyy-MM-dd');