GuidesBlog
GuidesAPI ReferenceDiscussions
GuidesBlogPlatform
Platform

Runtime Environment Configuration

Post-installation configuration steps that you need to perform before you can start using DBND.

Suggest Edits

Environments Overview

DBND provides out-of-the-box environments that you need to configure before you can start running your pipelines.
Out-of-the-box, DBND supports the following environment types:

  • Persistency - Local file system, AWS S3, Google Storage, Azure Blob Store, and HDFS
  • Spark - Local Spark, Amazon EMR, Google DataProc, Databricks, Qubole, and Livy
  • Docker Engine - Local Docker, AWS Batch, Kubernetes.
    You can also create custom environments and engines.

Main Configuration

• The environments parameter in the core section specifies a list of environments enabled and available for the project.
Possible values include local, gcp, aws, azure, and local - set by default.

[core]
environments = ['local', 'gcp']

The following describe the environment types supported in DBND:

Local

In the default local environment setup, the configuration works as follows:
A persistent metadata store for task inputs/outputs, metrics, and runtime execution information is set to the local file system under $DBND_HOME/data.
Python tasks run as processes on the local machine.
Spark tasks run locally by using the spark_submit command if you have local Spark in place.
Docker tasks run on your local Docker container.

[local] Configuration Section Parameter Reference

  • env_label - Set the environment type to be used. E.g. dev, int, prod.
  • production - This indicates that the environment is production.
  • conn_id - Set the cloud connection settings.
  • root - Determine the main data output location.
  • local_engine - Set which engine will be used for local execution
  • remote_engine - Set the remote engine for the execution of driver/tasks
  • submit_driver - Enable submitting driver to remote_engine.
  • submit_tasks - Enable submitting tasks to remote engine one by one.
  • spark_config - Determine the Spark Configuration settings
  • spark_engine - Set the cluster engine to be used. E.g. local, emr (aws), dataproc (gcp), etc.
  • hdfs - Set the Hdfs cluster configuration settings
  • beam_config - Set the Apache Beam configuration settings
  • beam_engine - Set the Apache Beam cluster engine. E.g. local or dataflow.
  • docker_engine - Set the Docker job engine, e.g. docker or aws_batch

Google Cloud Platform (GCP), out-of-the-box

The Spark engine is preset for Google DataProc. To set up a GCP environment, you will need to provide a GS bucket as a root path for the metadata store and the Airflow connection ID with cloud authentication info.
See Setting up GCP Environment..

Amazon Web Service (AWS), out-of-the-box

The Spark engine is preset for Amazon EMR. To set an AWS environment, you need to provide an S3 bucket as a root path for the metadata store and the Airflow connection ID with cloud authentication information.
See Setting Up an AWS Environment.

Microsoft Azure (Azure), out-of-the-box

The Spark engine is preset for Databricks. To set up an Azure environment, you need to provide an Azure Blob Store bucket as a root path for the metadata store and the Airflow connection ID with cloud authentication information.
See Setting Up an Azure Environment.

Custom environment

You can define a custom environment from scratch or inherit settings from an existing environment. You can create custom environments for managing dev/staging/production lifecycles, which normally involves switching data and execution locations.
See Setting up a Custom Environment (Extending Configurations).

To use out-of-the-box environments, set up one or more of the environments described in the referenced topics.

Updated 3 months ago

What’s Next