
Accelerating environmental initiatives with OS Climate


Evan Smith
Technical Content Manager
Starburst Data
Monica Miller
Developer Advocate
Starburst
Evan Smith
Technical Content Manager
Starburst Data
Monica Miller
Developer Advocate
Starburst
Share
Data mesh has many advantages for organizations that use it to change their business processes towards a decentralized model. Below we highlight 10 benefits and challenges of data mesh.
Data mesh is more efficient compared to traditional ETL pipelines using central IT Teams.
The image below shows a typical bottleneck created through a centralized IT team.
With data mesh, every business division and team is empowered to use the data-as-a-product methodology to solve problems and add value. This leverages the expertise of the whole organization to help drive insights, improving the self-sufficiency and agility of domains.
The owners of each domain know the context of their data best, and empowering them to interact with datasets directly eliminates many potential points of confusion between teams.
The image below shows how data mesh empowers individual domains.
Data mesh, using data-as-a-product, enables organizations to expose data to all domains. This empowers sharing and collaboration across teams. It also avoids duplication of effort and helps ensure agility.
The image below shows how different domains can share data-as-a-product, increasing agility.
Data mesh relies on strong governance to maintain the integrity of the datasets being shared. Security and compliance will remain at the forefront of concern.
The image below shows how the four pillars of data mesh work together to open data sharing across domains while ensuring strong governance. This will have significant impact on the enterprise who are interested in becoming data-driven business.
Data mesh is not a small change to the organizational structure of data management. Not all businesses are ready to take on a change of this nature. For this reason, it is important to keep the following points in mind.
Data mesh is an organizational approach that often helps solve business problems more quickly when adopted. To benefit from this, it is important to keep the business problems themselves in focus. This will help to determine the correct course of action for a given business.
To assist with this, it is usually a good idea to include a description of how data mesh will be adopted and the impact that is hoped to be achieved.
Data mesh requires a significant effort. Understanding the size and scope of this change is necessary to avoid unexpected difficulties in its implementation. As with any major change, if poorly defined, it can lead to problems and exacerbate underlying shortcomings in current business processes.
To help manage the risks of the project, a thorough, concrete, and realistic change management strategy must be developed before deploying data mesh. This must include attention to all of the four pillars, particularly federated computational governance as it sets the ground rules for the system as a whole.
Each business is unique, and so too are its needs concerning data mesh.
To succeed, the change management strategy must take into account many factors, including:
Like other major projects that impact the business at a core operational level, data mesh requires buy-in from all key stakeholders. At a minimum, this typically includes central IT teams, individual domains, and management. The plan should be communicated clearly to all teams, and the changes in their roles and responsibilities must be articulated directly.
Data mesh is not a one-size-fits-all solution. It is a framework for organizational change that offers greater agility in most circumstances. Organizations can adopt some of the principles of data mesh without others. For example, data mesh could be phased in at a department-by-department level. The risks and rewards of each level of adoption should be identified early in the project and communicated to all team members.
Download the free book by Zhamak Dehghani, Data Mesh: Delivering Data-Driven Value at Scale