Our series on one of the hottest topics in data management began with a general data mesh overview and continued with a focus on the first principle of data mesh: domain-driven ownership.
This blog will focus on the second pillar: data as a product.
Four data mesh principles
-
Domain-driven ownership
-
Data as a product
-
Self-service data infrastructure
-
Federated computational governance
Data mesh, simplified
A quick refresher: Data mesh is a social-technical framework for data management that assumes one of the primary challenges of managing analytical workloads with legacy architecture is knowledge. Repositioning the domain expert — as opposed to a centralized source — at the center of the data ecosystem is data mesh’s first principle: domain-driven ownership.
Treating data as a product is data mesh’s second principle — and it highlights the value we place on data as a strategic organizational asset. If data is to be owned by the domain, then the data mesh organization no longer treats data as a byproduct of operations but as a strategic foundation. In this data management framework, data products are the “architectural quantum,” as coined by the originator of data mesh, Zhamak Deghani.
The data product encapsulates and implements all the necessary behavior and structural components to process and share data as a product.
Source: ‘How To Design The Data Product Architecture,’ O’Reilly.
Data as a product, explained
At its essence, this principle is about unlocking reliable long-term analytics value and reducing friction. A data mesh organization puts domain experts in charge of the data — and then applies product thinking to ensure the data roadmap meets the accessibility, governance, and usability needs of the organization. That means domain owners in data mesh organizations treat data as a product.
When we think of products, we think of laptops and cars and smartphones. And data mesh organizations think of data the same way as the product managers of the latest gadget or the hottest line of sports cars.
At a data mesh company, data products get a vision and strategy, and a product roadmap that spans from idea to R&D, release, maintenance, and retirement. Domain owners apply lifecycle planning to data, as my colleague Alex T’Kint wrote in a recent blog.
Above all, the ‘data as a product’ principle ensures that data is always measured by the value it brings to the people who use it. And since the enterprise data mesh organization includes domain-driven ownership, the people who know the most about the data are in the best position as stewards of their data products.
“Data Products (and the Data Mesh within which they operate) make data easy to find, consume, share, and govern. And to deliver this value, our job as a practitioner is to make Data Products easy to build, deploy, secure, and manage.”
– Eric Broda,The Anatomy of a Data Product.’ August 19 2022
What success looks like
You can create a product by treating data as a product. Maybe you already have. But how do you know if your product has made an impact? How do you determine success?
At Collibra, we’re fortunate to be able to apply data mesh principles with our Collibra Data Intelligence Cloud. Our enterprise data catalog empowers analysts and business managers to quickly find, understand and access the data they need, when they need it. From our experience as data experts, and our focus on data intelligence, we understand that a successful data product should achieve these three goals:
- Usable
- Valuable
- Feasible
Usable, valuable, feasible – oh, my!
Be usable
For a data product to be successful it must be usable. Perhaps this goes without saying. But if data is not discoverable or understood, then your data product is not fulfilling its product goals.
Data needs to be discoverable and understood by decision makers for them to make effective decisions. Data must also be meaningful on its own so it can be used without having to correlate with other sources of information (which may not be available at the time of decision-making).
Tip: A good way to ensure your data product is usable is to facilitate discoverability and understanding with a Glossary or a data catalog. When you can house, analyze, and extract metadata from your data sets, you can provide your users with a central location for data.
Offer value
A data product must be valuable. Is yours? It may seem like a philosophical question, but if your data product doesn’t add value, then it might be time to deprecate it. Data products should be meaningful on their own, and provide even more insight when correlated with other data products.
Be feasible
Finally, what is feasible? Is your data available and accessible to the right people with the right tools?
If you’re producing a data product, you should be able to easily pull in the data sets you need. If you’re a data manager, you should know how to access your data and how to put it into an analytical tool that can help you make better decisions. If you’re a business user, then someone on your team should be able to access and use the data that’s relevant to their work so they can understand what’s happening in real-time.
Making an honest assessment of your progress against these goals — usable, valuable, feasible — will go a long way toward ensuring your creating a data mesh culture.
Calm in today’s stormy world
Today’s environments are complex. Modern networks. Modern application frameworks. Modern workforces. In this world, silos don’t work and centralization doesn’t work either.
The enterprise data mesh organization needs to make stronger connections between the engineers who enable analytics, the analysts who curate it, and the business leaders who leverage it to make decisions. This is at the heart of the concept of Deghani’s idea about “architectural quantum.” If we are to successfully treat data as a product — and become a data mesh organization — then data products must be incorporated into enterprise operational culture and workflows.
With data as a product as a core principle of the decentralized data mesh organization, the stakes become even greater for related data management measures, such as data quality. Part of the Collibra Data Intelligence Cloud, the Collibra Data Catalog offers a single point of engagement for discoverability and access to trusted, reliable data.
