Apache Airflow vs. Rivery vs. Stitch
ETL software comparison
Most businesses have data stored in a variety of locations, from in-house databases to SaaS platforms. To get a full picture of their finances and operations, they pull data from all those sources into a data warehouse or data lake and run analytics against it. But they don't want to build and maintain their own data pipelines.
Fortunately, it’s not necessary to code everything in-house. Here's an comparison of two such tools, head to head.
Apache Airflow is an open source project that lets developers orchestrate workflows to extract, transform, load, and store data.
Rivery is a cloud-based ETL platform.
Stitch Data Loader is a cloud-based platform for ETL — extract, transform, and load. More than 3,000 companies use Stitch to move billions of records every day from SaaS applications and databases into data warehouses and data lakes, where it can be analyzed with BI tools. Stitch is a Talend company and is part of the Talend Data Fabric.
|Orchestration, scheduling, workflows
|Data ingestion, ELT
|Data ingestion, ELT
|Only via plugins
|Full table; incremental by field timestamps
|Full table; incremental via change data capture or SELECT/replication keys
|Only via plugins
|More than 100
|Ability for customers to add new data sources
|Connects to data warehouses? Data lakes?
|Yes / Yes
|Yes / Yes
|Yes / Yes
|G2 customer satisfaction
|Free to download and use
|Requires a conversation with sales
|Options for self-service or talking with sales. Also available from the AWS store.
|Compliance, governance, and security certifications
|GDPR, SOC 2
|HIPAA, GDPR, SOC 2
|Yes, via plugins
|Yes, through Talend Data Fabric
|Free to use
|Annual or monthly contracts. No open source
|Month to month or annual contracts. Open source integrations
|Experimental REST API
|REST API, RIvery API for external embedding
|Import API, Stitch Connect API for integrating Stitch with other platforms, Singer open source project
Let's dive into some of the details of each platform.
Apache Airflow is a powerful tool for authoring, scheduling, and monitoring workflows as directed acyclic graphs (DAG) of tasks. A DAG is a topological representation of the way data flows within a system. Airflow manages execution dependencies among jobs (known as operators in Airflow parlance) in the DAG, and programmatically handles job failures, retries, and alerting. Developers can write Python code to transform data as an action in a workflow.
Rivery supports postload transformations defined in SQL via a feature called Logic Steps that can modify data within the target data warehouse.
Stitch is an ELT product. Within the pipeline, Stitch does only transformations that are required for compatibility with the destination, such as translating data types or denesting data when relevant. Stitch is part of Talend, which also provides tools for transforming data either within the data warehouse or via external processing engines such as Spark and MapReduce. Transformations can be defined in SQL, Python, Java, or via graphical user interface.
Try Stitch for free for 14 days
Each of these tools supports a variety of data sources and destinations.
Airflow orchestrates workflows to extract, transform, load, and store data. It run <b>tasks</b>, which are sets of activities, via <b>operators</b>, which are templates for tasks that can by Python functions or external scripts. Developers can create operators for any source or destination. In addition, Airflow supports plugins that implement operators and <b>hooks</b> — interfaces to external platforms. The Airflow community has built plugins for databases like MySQL and Microsoft SQL Server and SaaS platforms such as Salesforce, Stripe, and Facebook Ads.
Rivery supports more than 100 integrations, to databases, SaaS applications, and storage platforms. Rivery provides a Custom API feature that lets users send data to Rivery using scripts they build internally. It supports seven destinations: Amazon Redshift, Google BigQuery, and Snowflake data warehouses; Amazon S3 and Microsoft Azure Data Lake; and Azure Blog Storage and Google Cloud Storage.
Stitch supports more than 100 database and SaaS integrationsas data sources, and eight data warehouse and data lake destinations. Customers can contract with Stitch to build new sources, and anyone can add a new source to Stitch by developing it according to the standards laid out in Singer, an open source toolkit for writing scripts that move data. Singer integrations can be run independently, regardless of whether the user is a Stitch customer. Running Singer integrations on Stitch’s platform allows users to take advantage of Stitch's monitoring, scheduling, credential management, and autoscaling features.
Data integration tools can be complex, so vendors offer several ways to help their customers. Online documentation is the first resource users often turn to, and support teams can answer questions that aren't covered in the docs. Vendors of the more complicated tools may also offer training services.
The open source community provides Airflow support through a Slack community. Documentation includes quick start and how-to guides. Other than a tutorial on the Apache website there are no training resources.
Rivery provides support via Zendesk through a form on its website, and via email. Documentation is available from a link on the Rivery console. Rivery doesn't provide training services, but does offer video tutorials.
Stitch provides in-app chat support to all customers, and phone support is available for Enterprise customers. Support SLAs are available. Documentation is comprehensive and is open source — anyone can contribute additions and improvements or repurpose the content. Stitch does not provide training services.
Airflow is free and open source, licensed under Apache License 2.0.
Rivery provides a 14-day free trial. Pricing isn't disclosed, but it is "based on usage, which is calculated by the frequency in which your data streams are updated, and the frequency in which data transformations are executed."
Stitch has pricing that scales to fit a wide range of budgets and company sizes. All new users get an unlimited 14-day trial. Standard plans range from $100 to $1,250 per month depending on scale, with discounts for paying annually. Enterprise plans for larger organizations and mission-critical use cases can include custom features, data volumes, and service levels, and are priced individually.
Which tool is better overall? That's something every organization has to decide based on its unique requirements, but we can help you get started. Sign up now for a free trial of Stitch.