Data Mesh Principles: A New Approach Data Management

Blue Globe viewing from space at night with connections between cities. (World Map Courtesy of NASA:

Table Of Contents

Traditional data management poses challenges in processing vast and varied data. It also struggles to offer scalability, agility, and clear ownership. As a result, it often produces slow decision-making, data silos, and limited collaboration. Enter data mesh—a revolutionary approach designed to address these very issues.

Imagine data mesh as a symphony orchestra, where every instrument and musician plays a pivotal role. Data mesh emphasizes collaboration and coordination akin to the synchronicity in music. Within the data mesh framework, specialized teams act like musicians, each bringing unique expertise. And like a conductor, governance and coordination steer the data streams within the mesh, ensuring an easy flow of information.

At its core, data mesh champions the decentralization of data management. It emphasizes:

  • Shared data ownership so teams have direct control over their data
  • Subject-area approaches to address specific problems or areas of interest within the business
  • Support for independent teams by allowing teams to manage their data without central oversight

Data mesh decentralization spreads data ownership and decision-making across specialized teams, helping to foster a more responsive data ecosystem. With data mesh, companies can sidestep the bottlenecks of traditional systems and set the stage for enhanced teamwork, faster insights, and smoother operations.

Understanding traditional data management challenges

Traditional data management approaches can be challenging to use and manage. They frequently limit scalability, agility, and effectiveness.

Limitations of centralized data architectures

Centralized data architectures tend to limit enterprises through:

  • Lack of scalability: The inability to keep up with the ever-growing volume, variety, and velocity of data
  • Single points of failure: The entire network is at risk if the central system fails 
  • Data silos: Isolated data storage that restricts organizational access and hinders collaboration

Scalability issues and bottlenecks in data processing

Scalability issues in data processing are a headache for enterprises with massive amounts of data. When data volumes rise, traditional processing systems often struggle to meet demand. The sheer volume and speed at which data comes pouring in may lead to delays, performance hiccups, and even system failures. Throwing in more resources or hardware provides only temporary relief, and, many enterprises ultimately find the challenges of scaling too daunting.

Here are a few scalability issues businesses may encounter:

  • Scalability limitation: Constraints preventing systems from scaling to process more data
  • Scale-out challenges: Difficulties in expanding systems to process higher data loads
  • Resource exhaustion: Systems running out of critical resources like CPU, memory, or disk space
  • Throughput bottlenecks: A single limiting component restricts overall system throughput
  • Processing bottlenecks: Specific stages in the data pipeline slow the entire processing rate
  • Ingestion rate mismatch: Data intake rate exceeds processing capacity, causing data backup

Lack of ownership and accountability in data governance

Lack of clarity around data governance responsibilities is a major challenge for enterprises. It impacts their ability to maintain a robust data ecosystem and blurs the lines of responsibility for data professionals to use data.  

Let’s say a marketing team collects data for a campaign but notices some customer details don’t match. It’s hard to tell if the mistake came from the team entering the data, the group analyzing it, or the IT folks handling the database. Without knowing who’s responsible, fixing the problem gets messy and can cause delays and cost extra money.

Lack of clear ownership leads to confusion and inefficiencies. If no one’s in charge, important data tasks like checking quality, keeping private info safe, and controlling who gets access can get overlooked.

A centralized, automated data lineage system makes it simpler to spot and fix problems.  Approaching troubleshooting this way maintains data quality and ensures compliance. However, it’s still essential to have team members monitoring this technology. Lack of clear ownership, coupled with fragmented systems and procedures, jeopardizes the entire data infrastructure of an organization. Enterprises that contend with this challenge discover that it’s not operational; it’s strategic.

Data becomes fragmented across different systems, lacking proper documentation or metadata. Not only does fragmentation make data hard to find and access, but it also exposes organizations to data security risks. That’s why enterprises must establish clear ownership and accountability in data governance. Doing so lays a solid foundation for maintaining data integrity and building stakeholder trust.

How data mesh solves the problems of traditional data management

Traditional data management issues revolve around scalability, agility, and lack of clear ownership. Let’s see how data mesh solves these problems.


Data mesh lets organizations process and analyze data across multiple systems, fixing traditional scalability problems. Enterprises that confine processing and analysis to just a few systems are more prone to getting overwhelmed. Data mesh fixes this problem by allowing enterprises to manage and analyze data across distributed systems. Enhanced scalability ensures the system adapts to meet increasing demands. 


Agility is a remarkable benefit of data mesh principles. It propels enterprises to speed up data product development and iteration cycles. Agile organizations orchestrate rapid prototyping, testing, and iteration to refine their data products. Data mesh shortens the feedback loop and promotes continuous iteration so businesses stay ahead of the curve. 

Uber has successfully implemented data mesh to improve its data agility and decision-making. Before implementing data mesh, Uber used a centralized data structure. However, this setup made it hard for internal teams to access and share data. The architecture led to slow decision-making, poor communication, and data silos.

Under data mesh, teams own and manage the data they create. These teams can innovate and experiment with data without going through a central data team. It also makes it easier for teams to share data, which improves communication and collaboration. After adopting data mesh, Uber experienced faster decision-making and better communication. They also saw fewer data silos and a boost in innovation.

Ownership and accountability

Continuing the symphony analogy, ownership in the data mesh setup is like musicians in an orchestra. Just as every musician must play their part, a good data mesh system needs teams to really care for their data. By spreading out responsibilities for data, these teams, look after their data and ensure it is top-notch and usable. Their deep-rooted sense of ownership creates a lively, data-focused environment where teams proudly showcase their data together.

Introducing data mesh principles

Data mesh principles complement each other by addressing different aspects of data management. They work together to create a comprehensive approach to data management that empowers teams. Data mesh technology also promotes collaboration and ensures effectiveness and scalability. 

The most relevant data mesh principles are:

  • Decentralization
  • Subject-driven decentralized data architecture
  • Self-serve data infrastructure
  • Federated computational governance


In a data mesh system, decentralization shifts data ownership, accountability, and decision-making from a single centralized team to several subject-oriented teams. Rather than one group controlling all data decisions and governance, these teams manage their own specific data areas. 

Ownership in this context is about ensuring data quality, governance, and ease of use. It doesn’t pertain to where the data is physically stored. While subject teams validate the data’s relevance and accuracy, enterprises decide on infrastructure matters, such as data hosting locations.

Domain-driven decentralized data architecture

Instead of a centralized system, each data mesh team handles its own specific data, treating it like a unique product. The teams retain more control over their data, adjusting it to their needs and expertise. It emphasizes the importance of each team taking charge of their data’s quality and governance.

Gilead Sciences, a biopharmaceutical company, adopted data mesh to improve its drug discovery process. Before, Gilead’s centralized data system led to slow decisions, poor communication, and data silos. Relying upon this process made it difficult for teams to collaborate and share data, which slowed down the drug discovery process. 

By transitioning to data mesh, Gilead gave individual teams ownership and management of their data. Employing this decentralized framework allowed teams to innovate and experiment without interference from a centralized authority. It also made it easier for teams to collaborate and share data, which sped up the drug discovery process.

Self-serve data infrastructure

A self-serve data infrastructure provides teams with the tools and freedom to access, process, and analyze data without going through IT. The goal is to reduce bottlenecks, improve agility, and give teams more control over their data.

Federated computational governance

Imagine you have a giant puzzle you can’t solve by yourself. So, you invite your friends over to help you. Federated computational governance is a bit like puzzle-solving teamwork.

“Federated” means working together as a group, not just one person making decisions. “Computational governance” is like setting rules for how you all are going to work on the puzzle. You want to make sure everyone follows the same rules so things don’t get messy.

If the puzzle is a bunch of data you need to sort out, the “data mesh framework” is how you organize the puzzle pieces. Each team works on its part of the puzzle (or data), but they also need to talk to each other to make sure everything fits together.

So, “federated computational governance” means that different groups are working together and following specific rules to sort out and organize data in the correct way.

How to use data mesh principles

Enterprises that adopt data mesh principles must consider several key factors. The recipe for success involves: 

  • A collaborative culture that emphasizes knowledge-sharing
  • Ddomain teams with the resources necessary for data-driven decisions 
  • Robust data governance practices 
  • Scalable infrastructure 
  • Decentralized decision-making and accountability

To ensure success for enterprises adopting data mesh principles, they should adhere to the following four steps.

Step 1: Set up teams and data domains

Implementing a successful data mesh requires enterprises to form teams around specific topics and partition the data into distinct section. It also involves selecting experts for each topic and aligning them with the appropriate data sections. The process is akin to having violin players sit in one group and oboe players sit in another group within an orchestra. When you do this, each team becomes responsible for their data. Teams that do this well feel like they own and care for their data.

Step 2: Design for self-service

To embrace data mesh principles, enterprises should establish a self-serve data infrastructure tailored for data product development. Building an infrastructure that’s both scalable and adaptable ensures domain teams can autonomously access, process, and analyze data. Self-serve capabilities allow domain teams to create and remix their data products, spurring innovation and creativity.

Step 3: Don’t forget about governance

Implementing data mesh principles also requireseffective federated computational governance. Think of federated computational governance as the conductor of a symphony orchestra, synchronizing and coordinating the efforts of domain teams. By establishing collaborative decision-making processes and coordination mechanisms, enterprises can create a synchronized and cohesive approach to data governance. This process ensures every team plays their part in unison, collectively defining and enforcing data governance standards, data quality measures, and privacy controls throughout the data ecosystem.

Step 4: Watch out for pitfalls

Implementing data mesh principles comes with its own set of challenges and pitfalls. Resistance to change, especially when shifting from centralized to decentralized data management models, can feel like giving a dog a bath. It requires finesse, patience, and maybe even a little dog treat. Additional challenges include aligning domain teams on standards and governance practices, upholding consistent data quality across domains, and fostering effective communication among teams.

Organizations must proactively anticipate and address these challenges through careful planning and thoughtful strategies. When they do so, they foster a culture of openness and transparency and provide the necessary resources and support to successfully implement data mesh principles. Equipped with the required resources and support, enterprises can implement data mesh principles and meet any challenge head-on, leaving their competitors in awe.

Data mesh supports data productization

A successful data mesh implementation requires teams to work togeter. The data governance team ensures everyone is on the same page and working towards the same goal. The harmony of a symphony is similar to the synchronicity of data management, where everyone works together to achieve a common goal.

As we have seen, there are many benefits of data mesh principles for scalable and decentralized data management. Businesses can get more benefits from data mesh by working together, just like a symphony orchestra. Data mesh encourages teams to form and focus on specific areas of data, creating clear boundaries between data domains. Enterprises that embark on this journey ultimately achieve success in the data-driven economy.

Businesses seeking to leverage their data productization should turn to Revelate. Revelate’s data fulfillment platform helps businesses discover external data, making it easier to succeed in data productization. Data mesh principles support data productization by giving teams ownership of their data, which can lead to better data quality, collaboration, and agility. Revelate’s platform is designed to work with data mesh principles, making it a natural choice for businesses that want to adopt this approach to data management.

Unlock Your Data's Potential with Revelate

Revelate provides a suite of capabilities for data sharing and data commercialization for our customers to fully realize the value of their data. Harness the power of your data today!

Get Started