Temporal Tables in SQL Server and Azure SQL Database
Temporal table: to provide information about data in a table at any point in time.
On SQL Server from the 2016 version.
-- This statement creates a Temporal Table
CREATE TABLE dbo.ExampleTemporal
(
Id INT NOT NULL PRIMARY KEY CLUSTERED,
Col1 INT,
Col2 varchar(20),
SysStartTime datetime2 GENERATED ALWAYS AS ROW START NOT NULL,
SysEndTime datetime2 GENERATED ALWAYS AS ROW END NOT NULL,
PERIOD FOR SYSTEM_TIME (SysStartTime, SysEndTime) --Specifies which are the period column
)
WITH
(
SYSTEM_VERSIONING = ON (HISTORY_TABLE = dbo.ExampleHistory) --Table used to store the history
)
This is what you see on SSMS after the execution of the script:
As you can see, temporal tables are made of two tables:
the temporal table that keeps the current data just like your normal tables, with two additional fields: the period columns that represent the period of validity of the data for each row.
the history table which keeps the previous values of the data, with an entry per each change. Each change have a period of validity that can be used to retrieve value of a row for a period. Changes are stored automatically.
What happens to the two tables when you made an INSERT/UPDATE/DELETE?
When you INSERT a new row on a temporal table, the period of validity is set from the current time to the date 9999-12-31.
When you UPDATE a row, the previous row was stored in the history table (with the validity that starts with the time the record is created or updated previously and end now) and the data is updated in the temporal table (with the period of validity valid from now to 9999-12-31).
When you DELETE a row, the row is removed from the temporal table and stored in the history table with the correspondent period of validity which ends in the moment when the record was deleted.
--Insert an example
INSERT INTO dbo.ExampleTemporal (Id, Col1, Col2)
VALUES (1, 1, 'a')
--Update the example
UPDATE dbo.ExampleTemporal
SET Col1 = 2
WHERE Id = 1
--See only the current record
SELECT *
FROM dbo.ExampleTemporal
--Query the history table with all the old values
SELECT *
FROM dbo.ExampleHistory
WHERE Id = 1
Selecting the data a from temporal table is no different from selecting data in a normal table.
But you can use a few keywords in addition to the normal select statement to interact with the period of the data.
This keywords are:
FOR SYSTEM_TIME AS OF: used with a specific datetime -> AS OF datetime
FOR SYSTEM_TIME FROM TO: used with ranges of datetime -> FROM start TO end
FOR SYSTEM_TIME BETWEEN: used with ranges of datetime -> BETWEEN start AND end
FOR SYSTEM_TIME CONTAINED IN: used with ranges of datetime -> CONTAINED IN (start, end)
FOR SYSTEM_TIME ALL: all changes made to a record
--See only the current record
SELECT *
FROM dbo.ExampleTemporal
--See the record in a past state (different methods)
SELECT *
FROM dbo.ExampleTemporal
FOR SYSTEM_TIME AS OF '2022-08-08 16:22'
WHERE Id = 1
SELECT *
FROM dbo.ExampleTemporal
FOR SYSTEM_TIME BETWEEN '2022-08-05 16:22' AND '2022-08-09'
WHERE Id = 1
SELECT *
FROM dbo.ExampleTemporal
FOR SYSTEM_TIME FROM '2022-08-05 16:22' TO '2022-08-09'
WHERE Id = 1
--See the current and all the past versions of the record
SELECT *
FROM dbo.ExampleTemporal
FOR SYSTEM_TIME ALL
WHERE Id = 1
ORDER BY SysEndTime
With a temporal table, you can restore data back to a previous state using the FOR_SYSTEM_TIME AS OF clause. It restores both the updated and deleted records. You can use it for the entire table or only for part of it using the WHERE clause.
--Delete the record
DELETE FROM dbo.ExampleTemporal
WHERE Id = 1
--The record is gone
SELECT * FROM dbo.ExampleTemporal WHERE Id = 1
--Insert again the record from a past version of it
INSERT INTO dbo.ExampleTemporal (Id, Col1, Col2)
SELECT Id, Col1, Col2
FROM dbo.ExampleTemporal FOR SYSTEM_TIME AS OF '2022-08-08 16:30'
WHERE Id = 1
Subscribe to my newsletter
Read articles from Massimiliano Figini directly inside your inbox. Subscribe to the newsletter, and don't miss out.
Written by
Massimiliano Figini
Massimiliano Figini
data engineer • T-SQL developer • r and python • Power BI and tableau • machine learning enthusiast • data lover Follow back granted