Mirroring policy

Article
10/08/2024

The mirroring policy commands allow you to view, change, partition, and delete your table mirroring policy. They also provide a way to check the mirroring latency by reviewing the operations mirroring status.

Management commands

Use .show table policy mirroring command to show the current mirroring policy of the table.
Use .alter-merge table policy mirroring command to change the current mirroring policy.
Use .delete table policy mirroring command to soft-delete the current mirroring policy.
Use .show table mirroring operations command to check operations mirroring status.
Use .show table mirroring operations exported artifacts command to check operations exported artifacts status.
Use .show table mirroring operations failures to check operations mirroring failure status.

The policy object

The mirroring policy includes the following properties:

Property	Description	Values	Default
Format	The format of your mirrored files.	Valid value is `parquet`.	`parquet`
ConnectionStrings	An array of connection strings that help configure and establish connections. This value is autopopulated.
IsEnabled	Determines whether the mirroring policy is enabled. When the mirroring policy is disabled and set to `false`, the underlying mirroring data is retained in the database.	`true`, `false`, `null`.	`null`
Partitions	A comma-separated list of columns used to divide the data into smaller partitions.	See Partitions formatting.

Warning

Deleting or dropping the table mirroring policy will permanently delete the delta table in OneLake.

Note

When the mirroring policy is enabled, tables can't be renamed.

Partitions formatting

The partitions list is any combination of partition columns, specified using one of the forms shown in the following table.

Partition Type	Syntax	Notes
Virtual column	PartitionName `:` (`datetime` \| `string`)	Read more on Virtual columns.
String column value	PartitionName `:` `string` `=` ColumnName
String column value hash()	PartitionName `:` `long` `=` `hash(`ColumnName`,` Number`)`	The hash is modulo Number.
Truncated datetime column (value)	PartitionName `:` `datetime` `=` (`startofyear` \| `startofmonth` \| `startofweek` \| `startofday`) `(` ColumnName `)`	See documentation on startofyear, startofmonth, startofweek, or startofday functions.
Truncated Datetime Column Value `=` `bin` `(` ColumnName `,` TimeSpan `)`	Read more about the bin function.

Note

Each partition is represented as a separate column using the PartitionName listed in the Partitions list. PartitionName must be a case insensitive unique string, both among other partition names and the column names of the mirrored table.

Data types mapping

To ensure compatibility and optimize queries, ensure that your data types are properly mapped to the parquet data types.

Event house to Delta parquet data types mapping

Event house data types are mapped to Delta Parquet data types using the following rules:

Event house data type	Delta data type
`bool`	`boolean`
`datetime`	`timestamp OR date (for date-bound partition definitions)`
`dynamic`	`string`
`guid`	`string`
`int`	`integer`
`long`	`long`
`real`	`double`
`string`	`string`
`timespan`	`long`
`decimal`	`decimal(38,18)`

For more information on Event house data types, see Scalar data types.

Example policy

{
  "Format": "parquet",
  "IsEnabled": true,
  "Partitions": null,
}

This feature isn't supported.