Skip to main content

Firebolt configurations

Setting quote_columns​

To prevent a warning, make sure to explicitly set a value for quote_columns in your dbt_project.yml. See the doc on quote_columns for more information.

seeds:
  +quote_columns: false  #or `true` if you have csv column headers with spaces

Model configuration for fact tables​

A dbt model can be created as a Firebolt fact table and configured using the following syntax:

dbt_project.yml
models:
  <resource-path>:
    +materialized: table
    +table_type: fact
    +primary_index: [ <column-name>, ... ]
    +indexes:
      - type: aggregating
        key_column: [ <column-name>, ... ]
        aggregation: [ <agg-sql>, ... ]
      ...

Fact table configurations​

ConfigurationDescription
materializedHow the model will be materialized into Firebolt. Must be table to create a fact table.
table_typeWhether the materialized table will be a fact or dimension table.
primary_indexSets the primary index for the fact table using the inputted list of column names from the model. Required for fact tables.
indexesA list of aggregating indexes to create on the fact table.
typeSpecifies that the index is an aggregating index. Should be set to aggregating.
key_columnSets the grouping of the aggregating index using the inputted list of column names from the model.
aggregationSets the aggregations on the aggregating index using the inputted list of SQL agg expressions.

Example of a fact table with an aggregating index​

{{ config(
    materialized = "table",
    table_type = "fact",
    primary_index = "id",
    indexes = [
      {
        type: "aggregating",
        key_column: "order_id",
        aggregation: ["COUNT(DISTINCT status)", "AVG(customer_id)"]
      }
    ]
) }}

Model configuration for dimension tables​

A dbt model can be materialized as a Firebolt dimension table and configured using the following syntax:

dbt_project.yml
models:
  <resource-path>:
    +materialized: table
    +table_type: dimension
    ...

Dimension tables do not support aggregation indexes.

Dimension table configurations​

ConfigurationDescription
materializedHow the model will be materialized into Firebolt. Must be table to create a dimension table.
table_typeWhether the materialized table will be a fact or dimension table.

How aggregating indexes are named​

In dbt-firebolt, you do not provide names for aggregating indexes; they are named programmatically. dbt will generate index names using the following convention:

<table-name>__<key-column>__<index-type>_<unix-timestamp-at-execution>

For example, a join index could be named my_users__id__join_1633504263 and an aggregating index could be named my_orders__order_date__aggregating_1633504263.

Managing ingestion via external tables​

dbt-firebolt supports dbt's external tables feature, which allows dbt to manage the table ingestion process from S3 into Firebolt. This is an optional feature but can be highly convenient depending on your use case.

More information on using external tables including properly configuring IAM can be found in the Firebolt documentation.

Installation of external tables package​

To install and use dbt-external-tables with Firebolt, you must:

  1. Add this package to your packages.yml:

    packages:
      - package: dbt-labs/dbt_external_tables
        version: <version>

  2. Add these fields to your dbt_project.yml:

    dispatch:
      - macro_namespace: dbt_external_tables
        search_order: ['dbt', 'dbt_external_tables']

  3. Pull in the packages.yml dependencies by calling dbt deps.

Using external tables​

To use external tables, you must define a table as external in your dbt_project.yml file. Every external table must contain the fields url, type, and object_pattern. Note that the Firebolt external table specification requires fewer fields than what is specified in the dbt documentation.

In addition to specifying the columns, an external table may specify partitions. Partitions are not columns and they cannot have the same name as columns. To avoid YAML parsing errors, remember to encase string literals (such as the url and object_pattern values) in single quotation marks.

dbt_project.yml syntax for an external table​

sources:
  - name: firebolt_external
    schema: "{{ target.schema }}"
    loader: S3

    tables:
      - name: <table-name>
        external:
          url: 's3://<bucket_name>/'
          object_pattern: '<regex>'
          type: '<type>'
          credentials:
            aws_key_id: <key-id>
            aws_secret_key: <key-secret>
          object_pattern: '<regex>'
          compression: '<compression-type>'
          partitions:
            - name: <partition-name>
              data_type: <partition-type>
              regex: '<partition-definition-regex>'
          columns:
            - name: <column-name>
              data_type: <type>

aws_key_id and aws_secret_key are the credentails that allow Firebolt access to your S3 bucket. Learn how to set them up by following this guide. If your bucket is public these parameters are not necessary.

Running external tables​

The stage_external_sources macro is inherited from the dbt-external-tables package and is the primary point of entry when using thes package. It has two operational modes: standard and "full refresh."

# iterate through all source nodes, create if missing, refresh metadata
$ dbt run-operation stage_external_sources

# iterate through all source nodes, create or replace (no refresh command is required as data is fetched live from remote)
$ dbt run-operation stage_external_sources --vars "ext_full_refresh: true"

Incremental models​

The incremental_strategy configuration controls how dbt builds incremental models. Firebolt currently supports append, insert_overwrite and delete+insert configuration. You can specify incremental_strategy in dbt_project.yml or within a model file's config() block. The append configuration is the default. Specifying this configuration is optional.

The append strategy performs an INSERT INTO statement with all the new data based on the model definition. This strategy doesn't update or delete existing rows, so if you do not filter the data to the most recent records only, it is likely that duplicate records will be inserted.

Example source code:

{{ config(
   materialized = 'incremental',
   incremental_strategy='append'
) }}

/* All rows returned by this query will be appended to the existing model */


select * from {{ ref('raw_orders') }}
{% if is_incremental() %}
   where order_date > (select max(order_date) from {{ this }})
{% endif %}

Example run code:

CREATE DIMENSION TABLE IF NOT EXISTS orders__dbt_tmp AS
SELECT * FROM raw_orders
WHERE order_date > (SELECT MAX(order_date) FROM orders);

INSERT INTO orders VALUES ([columns])
SELECT ([columns])
FROM orders__dbt_tmp;

Seeds behavior​

When running the dbt seed command we perform a DROP CASCADE operation instead of TRUNCATE.

Practice​

You can look at our modified version of the jaffle_shop, jaffle_shop_firebolt, to see how indexes, as well as external tables, can be set or clone and execute the commands listed in the README.md

0
Edit this page