Modifying Data

QuestDB is a timeseries database optimized to append data.

For best performance, design your application to avoid having to frequently edit existing records.

UPDATE and DELETE statements will be included in upcoming QuestDB versions, but even once they are made available these will be intended for correcting data that was inserted incorrectly or should have never been inserted in the first place (for example as part of data administration tasks).

These are three alternatives to UPDATE and DELETE you may consider:

  • Append newest state: Insert a newer state to replace an older one: This has the added advantage that you can query back in time to a previous state. It is also the basis of organizing data for bi-temporality.

  • Replace a table: Create a new table with the new data you need, drop the old one and rename.

  • Delete by dropping partitions: Create your timeseries tables with partitions, then delete the ones you no longer need.

Append newest state#

Using the timestamp field#

Here's a worked example using the timestamp column:

CREATE TABLE takeaway_order (
ts TIMESTAMP,
id SYMBOL,
status SYMBOL)
timestamp(ts);
INSERT INTO takeaway_order VALUES (now(), 'order1', 'placed');
INSERT INTO takeaway_order VALUES (now(), 'order2', 'placed');
INSERT INTO takeaway_order VALUES (now(), 'order1', 'cooking');
INSERT INTO takeaway_order VALUES (now(), 'order1', 'in-transit');
INSERT INTO takeaway_order VALUES (now(), 'order1', 'arrived');
INSERT INTO takeaway_order VALUES (now(), 'order3', 'placed');
INSERT INTO takeaway_order VALUES (now(), 'order3', 'cooking');
INSERT INTO takeaway_order VALUES (now(), 'order3', 'in-transit');

We join the latest timestamp of an order id against the rest of the data to obtain full details.

WITH
ts_takeaway_order AS (
SELECT
max(ts) AS ts,
id
FROM
takeaway_order GROUP BY id)
SELECT
o.*
FROM
ts_takeaway_order ts_o
INNER JOIN 'takeaway_order' o
ON ts_o.ts = o.ts

This results in the latest state for each order:

timestamp tsid symbolstatus symbol
2022-04-07T15:33:43.944922Zorder1arrived
2022-04-07T15:33:37.370694Zorder2placed
2022-04-07T15:33:50.829323Zorder3in-transit

Using dedicated fields#

If timestamps don't work for you here, you can also use an extra integer column called version, an extra boolean deleted column or similar.

Replace Table#

Another alternative is to:

  • Select only the data you want from an existing table into a new temporary one.
  • Drop the original table.
  • Rename the temporary table to the original table's name.
CREATE TABLE mytable_copy AS (
SELECT * FROM mytable WHERE column_value != 42
) TIMESTAMP(ts) PARTITION BY DAY;
DROP TABLE mytable;
RENAME table mytable_copy TO mytable;

Delete by Dropping Partitions#

When you create tables with a timestamp, you may organise them into partitions using the CREATE TABLE .. PARTITION BY SQL statement.

You may then use the ALTER TABLE DROP PARTITION SQL statement to drop partitions you no longer need.