Skip to main content

Add snapshots to your DAG

What are snapshots?

Analysts often need to "look back in time" at previous data states in their mutable tables. While some source data systems are built in a way that makes accessing historical data possible, this is not always the case. dbt provides a mechanism, snapshots, which records changes to a mutable table over time.

Snapshots implement type-2 Slowly Changing Dimensions over mutable source tables. These Slowly Changing Dimensions (or SCDs) identify how a row in a table changes over time. Imagine you have an orders table where the status field can be overwritten as the order is processed.

idstatusupdated_at
1pending2019-01-01

Now, imagine that the order goes from "pending" to "shipped". That same record will now look like:

idstatusupdated_at
1shipped2019-01-02

This order is now in the "shipped" state, but we've lost the information about when the order was last in the "pending" state. This makes it difficult (or impossible) to analyze how long it took for an order to ship. dbt can "snapshot" these changes to help you understand how values in a row change over time. Here's an example of a snapshot table for the previous example:

idstatusupdated_atdbt_valid_fromdbt_valid_to
1pending2019-01-012019-01-012019-01-02
1shipped2019-01-022019-01-02null

In dbt, snapshots are select statements, defined within a snapshot block in a .sql file (typically in your snapshots directory). You'll also need to configure your snapshot to tell dbt how to detect record changes.

snapshots/orders_snapshot.sql
{% snapshot orders_snapshot %}

{{
config(
target_database='analytics',
target_schema='snapshots',
unique_key='id',

strategy='timestamp',
updated_at='updated_at',
)
}}

select * from {{ source('jaffle_shop', 'orders') }}

{% endsnapshot %}
Preview or Compile Snapshots in IDE

It is not possible to "preview data" or "compile sql" for snapshots in dbt Cloud. Instead, run the dbt snapshot command in the IDE by completing the following steps.

When you run the dbt snapshot command:

  • On the first run: dbt will create the initial snapshot table — this will be the result set of your select statement, with additional columns including dbt_valid_from and dbt_valid_to. All records will have a dbt_valid_to = null.
  • On subsequent runs: dbt will check which records have changed or if any new records have been created:
    • The dbt_valid_to column will be updated for any existing records that have changed
    • The updated record and any new records will be inserted into the snapshot table. These records will now have dbt_valid_to = null

Snapshots can be referenced in downstream models the same way as referencing models — by using the ref function.

Example

To add a snapshot to your project:

  1. Create a file in your snapshots directory with a .sql file extension, e.g. snapshots/orders.sql
  2. Use a snapshot block to define the start and end of a snapshot:
snapshots/orders_snapshot.sql
{% snapshot orders_snapshot %}

{% endsnapshot %}
  1. Write a select statement within the snapshot block (tips for writing a good snapshot query are below). This select statement defines the results that you want to snapshot over time. You can use sources and refs here.
snapshots/orders_snapshot.sql
{% snapshot orders_snapshot %}

select * from {{ source('jaffle_shop', 'orders') }}

{% endsnapshot %}
  1. Check whether the result set of your query includes a reliable timestamp column that indicates when a record was last updated. For our example, the updated_at column reliably indicates record changes, so we can use the timestamp strategy. If your query result set does not have a reliable timestamp, you'll need to instead use the check strategy — more details on this below.

  2. Add configurations to your snapshot using a config block (more details below). You can also configure your snapshot from your dbt_project.yml file (docs).

snapshots/orders_snapshot.sql
{% snapshot orders_snapshot %}

{{
config(
target_database='analytics',
target_schema='snapshots',
unique_key='id',

strategy='timestamp',
updated_at='updated_at',
)
}}

select * from {{ source('jaffle_shop', 'orders') }}

{% endsnapshot %}
  1. Run the dbt snapshot command — for our example a new table will be created at analytics.snapshots.orders_snapshot. You can change the target_database configuration, the target_schema configuration and the name of the snapshot (as defined in {% snapshot .. %}) will change how dbt names this table.
$ dbt snapshot
Running with dbt=0.16.0

15:07:36 | Concurrency: 8 threads (target='dev')
15:07:36 |
15:07:36 | 1 of 1 START snapshot snapshots.orders_snapshot...... [RUN]
15:07:36 | 1 of 1 OK snapshot snapshots.orders_snapshot..........[SELECT 3 in 1.82s]
15:07:36 |
15:07:36 | Finished running 1 snapshots in 0.68s.

Completed successfully

Done. PASS=2 ERROR=0 SKIP=0 TOTAL=1
  1. Inspect the results by selecting from the table dbt created. After the first run, you should see the results of your query, plus the snapshot meta fields as described below.

  2. Run the snapshot command again, and inspect the results. If any records have been updated, the snapshot should reflect this.

  3. Select from the snapshot in downstream models using the ref function.

models/changed_orders.sql
select * from {{ ref('orders_snapshot') }}
  1. Schedule the snapshot command to run regularly — snapshots are only useful if you run them frequently.

Detecting row changes

Snapshot "strategies" define how dbt knows if a row has changed. There are two strategies built-in to dbt — timestamp and check.

The timestamp strategy uses an updated_at field to determine if a row has changed. If the configured updated_at column for a row is more recent than the last time the snapshot ran, then dbt will invalidate the old record and record the new one. If the timestamps are unchanged, then dbt will not take any action.

The timestamp strategy requires the following configurations:

ConfigDescriptionExample
updated_atA column which represents when the source row was last updatedupdated_at

Example usage:

snapshots/orders_snapshot_timestamp.sql
{% snapshot orders_snapshot_timestamp %}

{{
config(
target_schema='snapshots',
strategy='timestamp',
unique_key='id',
updated_at='updated_at',
)
}}

select * from {{ source('jaffle_shop', 'orders') }}

{% endsnapshot %}

Check strategy

The check strategy is useful for tables which do not have a reliable updated_at column. This strategy works by comparing a list of columns between their current and historical values. If any of these columns have changed, then dbt will invalidate the old record and record the new one. If the column values are identical, then dbt will not take any action.

The check strategy requires the following configurations:

ConfigDescriptionExample
check_colsA list of columns to check for changes, or all to check all columns["name", "email"]
check_cols = 'all'

The check snapshot strategy can be configured to track changes to all columns by supplying check_cols = 'all'. It is better to explicitly enumerate the columns that you want to check. Consider using a surrogate key to condense many columns into a single column.

Example Usage

snapshots/orders_snapshot_check.sql
{% snapshot orders_snapshot_check %}

{{
config(
target_schema='snapshots',
strategy='check',
unique_key='id',
check_cols=['status', 'is_cancelled'],
)
}}

select * from {{ source('jaffle_shop', 'orders') }}

{% endsnapshot %}

Hard deletes (opt-in)

Rows that are deleted from the source query are not invalidated by default. With the config option invalidate_hard_deletes, dbt can track rows that no longer exist. This is done by left joining the snapshot table with the source table, and filtering the rows that are still valid at that point, but no longer can be found in the source table. dbt_valid_to will be set to the current snapshot time.

This configuration is not a different strategy as described above, but is an additional opt-in feature. It is not enabled by default since it alters the previous behavior.

For this configuration to work with the timestamp strategy, the configured updated_at column must be of timestamp type. Otherwise, queries will fail due to mixing data types.

Example Usage

snapshots/orders_snapshot_hard_delete.sql
{% snapshot orders_snapshot_hard_delete %}

{{
config(
target_schema='snapshots',
strategy='timestamp',
unique_key='id',
updated_at='updated_at',
invalidate_hard_deletes=True,
)
}}

select * from {{ source('jaffle_shop', 'orders') }}

{% endsnapshot %}

Configuring snapshots

Snapshot configurations

There are a number of snapshot-specific configurations:

ConfigDescriptionRequired?Example
target_databaseThe database that dbt should render the snapshot table intoNoanalytics
target_schemaThe schema that dbt should render the snapshot table intoYessnapshots
strategyThe snapshot strategy to use. One of timestamp or checkYestimestamp
unique_keyA primary key column or expression for the recordYesid
check_colsIf using the check strategy, then the columns to checkOnly if using the check strategy["status"]
updated_atIf using the timestamp strategy, the timestamp column to compareOnly if using the timestamp strategyupdated_at
invalidate_hard_deletesFind hard deleted records in source, and set dbt_valid_to current time if no longer existsNoTrue

A number of other configurations are also supported (e.g. tags and post-hook), check out the full list here.

Snapshots can be configured from both your dbt_project.yml file and a config block, check out the configuration docs for more information.

Note: BigQuery users can use target_project and target_dataset as aliases for target_database and target_schema, respectively.

Configuration best practices

Use the timestamp strategy where possible

This strategy handles column additions and deletions better than the check strategy.

Ensure your unique key is really unique

The unique key is used by dbt to match rows up, so it's extremely important to make sure this key is actually unique! If you're snapshotting a source, I'd recommend adding a uniqueness test to your source (example).

Use a target_schema that is separate to your analytics schema

Snapshots cannot be rebuilt. As such, it's a good idea to put snapshots in a separate schema so end users know they are special. From there, you may want to set different privileges on your snapshots compared to your models, and even run them as a different user (or role, depending on your warehouse) to make it very difficult to drop a snapshot unless you really want to.

Snapshot query best practices

Snapshot source data.

Your models should then select from these snapshots, treating them like regular data sources. As much as possible, snapshot your source data in its raw form and use downstream models to clean up the data

Use the source function in your query.

This helps when understanding data lineage in your project.

Include as many columns as possible.

In fact, go for select * if performance permits! Even if a column doesn't feel useful at the moment, it might be better to snapshot it in case it becomes useful – after all, you won't be able to recreate the column later.

Avoid joins in your snapshot query.

Joins can make it difficult to build a reliable updated_at timestamp. Instead, snapshot the two tables separately, and join them in downstream models.

Limit the amount of transformation in your query.

If you apply business logic in a snapshot query, and this logic changes in the future, it can be impossible (or, at least, very difficult) to apply the change in logic to your snapshots.

Basically – keep your query as simple as possible! Some reasonable exceptions to these recommendations include:

  • Selecting specific columns if the table is wide.
  • Doing light transformation to get data into a reasonable shape, for example, unpacking a JSON blob to flatten your source data into columns.

Snapshot meta-fields

Snapshot tables will be created as a clone of your source dataset, plus some additional meta-fields*.

FieldMeaningUsage
dbt_valid_fromThe timestamp when this snapshot row was first insertedThis column can be used to order the different "versions" of a record.
dbt_valid_toThe timestamp when this row became invalidated.The most recent snapshot record will have dbt_valid_to set to null.
dbt_scd_idA unique key generated for each snapshotted record.This is used internally by dbt
dbt_updated_atThe updated_at timestamp of the source record when this snapshot row was inserted.This is used internally by dbt

*The timestamps used for each column are subtly different depending on the strategy you use:

For the timestamp strategy, the configured updated_at column is used to populate the dbt_valid_from, dbt_valid_to and dbt_updated_at columns.

Details for the timestamp strategy

Snapshot query results at 2019-01-01 11:00

idstatusupdated_at
1pending2019-01-01 10:47

Snapshot results (note that 11:00 is not used anywhere):

idstatusupdated_atdbt_valid_fromdbt_valid_todbt_updated_at
1pending2019-01-01 10:472019-01-01 10:472019-01-01 10:47

Query results at 2019-01-01 11:30:

idstatusupdated_at
1shipped2019-01-01 11:05

Snapshot results (note that 11:30 is not used anywhere):

idstatusupdated_atdbt_valid_fromdbt_valid_todbt_updated_at
1pending2019-01-01 10:472019-01-01 10:472019-01-01 11:052019-01-01 10:47
1shipped2019-01-01 11:052019-01-01 11:052019-01-01 11:05

For the check strategy, the current timestamp is used to populate each column. If configured, the check strategy uses the updated_at column instead, as with the timestamp strategy.

Details for the check strategy

Snapshot query results at 2019-01-01 11:00

idstatus
1pending

Snapshot results:

idstatusdbt_valid_fromdbt_valid_todbt_updated_at
1pending2019-01-01 11:002019-01-01 11:00

Query results at 2019-01-01 11:30:

idstatus
1shipped

Snapshot results:

idstatusdbt_valid_fromdbt_valid_todbt_updated_at
1pending2019-01-01 11:002019-01-01 11:302019-01-01 11:00
1shipped2019-01-01 11:302019-01-01 11:30

FAQs

0