Building Data Mesh Architectures with Dataflow on Google Cloud

By Dishani Sen

Hi again, data enthusiasts! Welcome to an exciting blog where we will explore the world of data mesh architectures and how they help solve data-related challenges for businesses. In this blog, we will delve into the concept of data mesh, the problems it addresses, and the role of Dataflow on Google Cloud Platform (GCP) in implementing data mesh architectures. We will also demonstrate how Dataflow seamlessly integrates with data mesh to empower organisations with efficient data management and analysis.

The Challenge: Exploding Data Growth and Extracting Value

Data is growing exponentially, with production increasing approximately 10x over the past eight years. However, many companies are struggling to extract meaningful and measurable value from their data investments. Around 70% of businesses worldwide are facing difficulties in realizing the true potential of their data. Bridging this gap, known as the "data value bridge," has become a top priority for organisations.

The Data Mesh Solution: Decentralised Data Ownership with Centralised Provisioning

To address the data value bridge, businesses have several options, including centralised data lakes or data warehouses, adopting a data mesh architecture, or combining both approaches. In this blog, we will focus on data mesh architectures, which uniquely combine decentralised data ownership with centralised provisioning of infrastructure, governance policies, and data management best practices. This combination leads to faster time-to-value and increased agility in responding to external changes.

Data Mesh is a relatively new architectural approach to managing data in a decentralised and domain-oriented manner. It was introduced by Zhamak Dehghani, a ThoughtWorks principal consultant, in 2019. The concept of Data Mesh is based on the idea of treating data as a product and decentralising its ownership and governance across various domains within an organisation.
Traditionally, data management in organisations has been centralised, with a single data team responsible for all data-related tasks, including data storage, processing, and governance. However, as organisations grow and generate more data, this centralised approach can lead to challenges, such as bottlenecks in data access, delayed data delivery, and increased complexities in data ownership.

Data Mesh Building Blocks: Treating Data as a Product

At the core of the data mesh architecture is the concept of treating data as a product. The architecture begins by creating logical groups or domains, structured around the unique context of a business. These domains are empowered to produce and share their data products in a decentralised manner. Simultaneously, a centralised approach provides the necessary underlying technology infrastructure, governance policies, and data management best practices. The result is interoperability across data products and a standardised way to monitor and manage the data estate.

Dataflow: Empowering Data Mesh Architecture on GCP

Google Cloud's unified analytics solution, Dataflow, plays a pivotal role in implementing data mesh architectures. Let's explore how Dataflow supports the core building blocks of the data mesh:

  1. Data Products: Dataflow enables the creation of batch and streaming pipelines for ingesting and enriching domain data, transforming it into valuable data products.
  2. Decentralised Data Ownership: Dataflow supports multiple programming languages (Java, Go, and Python) and provides a variety of connectors, allowing different domains to create pipelines that best suit their needs.
  3. Self-Service Capabilities: Dataflow offers ready-to-use templates for quick start-up and simplified management through features like horizontal and vertical auto-scaling.
  4. Governance and Security: Dataflow seamlessly integrates with Google Cloud's Identity and Access Management (IAM) and advanced networking solutions, enabling federated governance for data mesh architectures.

Streaming Data Products: Empowering Real-time Insights

Dataflow plays a crucial role in processing streaming data for data mesh architectures. For instance, consider a multinational organisation's website managed by a team responsible for various regional domains. These teams may need real-time insights into web traffic changes caused by ad placements or external events. Dataflow is utilised to parse website click payloads, enrich the data with geolocation information, save it in BigQuery, and publish aggregated web traffic to Pub/Sub topics.

Consumers can discover these data products in the data catalog, request access, and create subscriptions to receive the relevant data. Dataflow ensures smooth real-time processing and enables consumers to further process streaming data in their own pipelines.

A Technical Guide to Getting Started with Google Dataflow on GCP

Google Dataflow is a fully managed service that provides a unified programming model for both batch and streaming data processing. In this guide, we'll walk you through the process of setting up Google Dataflow and creating basic data processing pipelines. Let's dive in!

Step 1: Create a Google Cloud Project and Enable Dataflow API

  1. Sign in to your Google account and navigate to the Google Cloud Console (
  2. Create a new project or select an existing one.
  3. In the Google Cloud Console, click on the "Navigation menu" > "Dataflow."
  4. If the Dataflow API is not enabled for your project, click on the "Enable" button to enable it.

Step 2: Set up Google Cloud SDK and Local Development Environment

  1. Install the Google Cloud SDK on your local machine (
  2. Open a terminal or command prompt and run gcloud init to authenticate and configure your project settings.

Step 3: Write a Simple Batch Data Processing Pipeline

  1. Create a new Python file (e.g., using your code editor.

Step 4: Run the Batch Data Processing Pipeline Locally

  1. In the terminal, navigate to the directory containing the file.
  2. Run the pipeline with the following command python


Step 3: Write a Simple Streaming Data Processing Pipeline

  1. Create a new Python file (e.g., using a  code editor.

Step 4: Run the Streaming Data Processing Pipeline Locally

  1. In the terminal, navigate to the directory containing the file.
  2. Run the pipeline with the following command python


Data mesh architectures offer a powerful solution to bridge the gap between data production and value extraction. Google Cloud's Dataflow perfectly complements data mesh by empowering organisations with the tools and capabilities needed to create data products, support decentralised data ownership, and enable real-time insights through streaming data products. By combining the strengths of data mesh and Dataflow, businesses can unlock the true potential of their data investments and drive data democratisation across the organisation.