Wondering about Data Mesh? What should you be considering when moving on from a central data platform architecture to a decentralized domain-oriented self-serve design? Check out what Anders Boje Hertz, Head of Data & AI at INTELLISHORE has to say.
2. Anders Boje Hertz
Head of AI & Data Platforms at INTELLISHORE
anders.hertz@Intellishore.dk
https://www.linkedin.com/in/andersboje/
3. We focus on providing the next generation of consultancy for the digital age: a true passion for
data, a combination of strategy advisors and native technologists, and a dedication to
delivering actionable insights and cultural change
45+
Employees
In 2013
It all started
3,500 employees trained
In analytics solutions developed by
Intellishore
200+ analytics
projects completed
34.8%
Top-line growth in 2020 &
29.7% in 2019
4. AND A COMPREHENSIVE TOOLKIT TO SUPPORT THE DIGITAL TRANSFORMATION
| NON-EXHAUSTIVE |
IGNITING DATA – ENGAGING PEOPLE | 2021
ENGAGING PEOPLE
IGNITING
DATA
Sources & Platforms
Industry
Automation
Decision Maturity
Learning Platform Visual Analytics
Algorithmic Science
User Journey
Strategizing | Data & AI
5. Defining a shared action plan for all
elements of transformation: business,
technical and organizational
001
Establishing the setup to continuously
validate, access, explore and expand the
foundation of quality data
002
Turning the insights from the data into accessible
reports and tools that can be used in day-to-day
operations
003
Building up the organization’s capabilities and
ways of working to drive data literacy and usage
004
Ensuring that data scientists, engineers and their
analyses are fully engaged and integrated with the day-
to-day business
005
TRANSFORMATIONS ARE DRIVEN BY HOW FAST COMPANIES MASTER 5 DISCIPLINES
IGNITING DATA – ENGAGING PEOPLE | 2020
9. Data Mesh
A decentralized sociotechnical
approach in managing and
accessing analytical data at scale.
10. IGNITING DATA – ENGAGING PEOPLE | 2021
Domain-oriented
decentralised data
ownership and architecture
Data as a
product
Self-serve data
infrastructure as
a platform
Federated
computational
governance
THE FOUR PRINCIPLES OF DISTRIBUTED ARCHITECTURE
11. IGNITING DATA – ENGAGING PEOPLE | 2021
Domains aligned
with the origin of
data
Domains aligned
with shared aggregates
Domains aligned with
the consumption
DECOMPOSE DATA AROUND DOMAINS
Distribute the ownership
12. IGNITING DATA – ENGAGING PEOPLE | 2021
Domain Data
Product Owner
L
T
E
Domain Data
Product
SERVE DATA AS A PRODUCT
Delight the consumer with ease of data discovery and use
13. IGNITING DATA – ENGAGING PEOPLE | 2021
ENABLE AUTONOMY
Abstract technical complexity in self-serve data infrastructure
Data | ML Infrastructure as a Platform
Data Infra Team
14. IGNITING DATA – ENGAGING PEOPLE | 2021
BUILD AN ECOSYSTEM
Create a federated and global governance
Data Infra as a Platform
Global Governance| Open Standards
15. IGNITING DATA – ENGAGING PEOPLE | 2021
Federated Computational
Governance
Apps
Multi-plane Data
Platforms
(Transactional Data, Code)
Domain’s App Devs
Data Platform Teams
Micro
Service
Legacy
App
Data Product as Architecture
Quantum
(Analytical Data, Meta-data, Code /
Pipeline, Policy sidecar)
Domain’s Data Product Devs and
Data
Product Owner
Domain representative
Domain DP owners
22. Data Mesh
A decentralized sociotechnical
approach in managing and
accessing analytical data at scale.
23. IGNITING DATA – ENGAGING PEOPLE | 2021
While data (for most) may not be a “product” in a
strictly economic sense, it is still the life-blood of
organizational decision making, and should not be
allowed to become a by-product.
Serge Gershkovich
24. REFERENCES
Dehghani, Z. (2020). Data Mesh Principles and Logical Architecture.
Retrieved from Martin Fowler: https://martinfowler.com/
Moses, B. (2022). Data Mesh 101: Everything You Need To Know to Get Started.
Retrieved from Monte Carlo Data: https://www.montecarlodata.com/blog-data-mesh-101-everything-you-need-to-know-to-get-started/
Thoughtworks. (2022). Introduction to Data Mesh A principled approach.
Retrieved from Thoughtworks: https://www.thoughtworks.com/what-we-do/data-and-ai/data-mesh
Data Mesh Learning. (2022). Intro to Data Mesh.
Retrieved from Data Mesh Learning: https://datameshlearning.com/
Dehghani, Z. (2019). How to Move Beyond a Monolithic Data Lake to a Distributed Data Mesh.
Retrieved from Martin Fowler: https://martinfowler.com/articles/data-monolith-to-mesh.html
Gershkovich, S. (2022). Data Mesh: overhyped, misunderstood, and useful!
Retrieved from Medium: https://medium.com/sqldbm/data-mesh-overhyped-misunderstood-and-useful-e65c60ba6643
27. Anders Boje Hertz
Head of AI & Data Platforms at INTELLISHORE
anders.hertz@Intellishore.dk
https://www.linkedin.com/in/andersboje/
Editor's Notes
Data is always own by a specific domain in the business.Access to that data I decentralized
A team is a business oriented technology team.
Data Mesh – we are extending the capabilities and also the accountability of those teams to serve and share data for analytical puporses and also embed ML and analytics in the domain.
Removing every data middelman
EXIT
Some other problem may arise, where you could imaging that this could create these data siloing, right?
A sales team have all the data they need to optimize processing my sales and I have no incentive to share that data with anybody else.
Beacuase I don’t care about their needs really.
So the second principle is “data as a product” that tries to address is the data silings.
Data is consided as a product by each team that publish the data.
Product thinking about that data.
It is not a asset we collect but it’s a product that we share and we are accountable of the experience for the analysis using the data.
Accountability is the key
The team is wholly responsible for it
Its quality
Its representation
Its cohesiveness
Like it was a thing they where sculpting something out of glass and polishing it and putting in at the storefront. They just want that thing to look nice and usefull.
Data is available everywhere and self-serve anywhere in the company.
Now I hear what you’re saying. You’re thinking about governance, that still is a thing but in principle, these data products are published and there are available everywhere.If you are producing sales report for at sales forecast for a company in Germany, you can find and source all of the data you need to drive that report. Getting that data from all that data where it lives to some database you have control of.
Address the cost and feasibility problem, that arise form this decentralized ownership of data prodcuts.
Look at infrastructure with a higher level of abstraction of complexity of data infrastructure then what we have to today.
So we can lower the cognitive loads of these teams so that genealists, ap developers, programmers, people we have in or organization are able to do data work, are able to create those analytics, insights and create and share data products
Last principle is about of doing all that with out compromising security, privacy, making sure that data is discoverable, when we blend different domains, we can still make sense of who the customer is.
https://www.youtube.com/watch?v=zfFyE3xmJ7I&ab_channel=Confluent
For any of these operations to be possible, a data mesh implementation requires a governance model that embraces decentralization and domain self-sovereignty, interoperability through global standardization, a dynamic topology and most importantly automated execution of decisions by the platform.
a “common ground” for the whole platform where all data products conform to a shared set of rules, where necessary while leaving enough space for autonomous decision-making.
The idea is to localize decisions as close to the source as possible while keeping interoperability and integration standards at a global level, so the mesh components can be easily integrated. In a data mesh, tools can be used to enforce global policies such as GDPR enforcement or access management and also local policies where each domain sets its own policies for their data products such as access control or data retention.
Last principle is about of doing all that with out compromising security, privacy, making sure that data is discoverable, when we blend different domains, we can still make sense of who the customer is.
https://www.youtube.com/watch?v=zfFyE3xmJ7I&ab_channel=Confluent
For any of these operations to be possible, a data mesh implementation requires a governance model that embraces decentralization and domain self-sovereignty, interoperability through global standardization, a dynamic topology and most importantly automated execution of decisions by the platform.
a “common ground” for the whole platform where all data products conform to a shared set of rules, where necessary while leaving enough space for autonomous decision-making.
The idea is to localize decisions as close to the source as possible while keeping interoperability and integration standards at a global level, so the mesh components can be easily integrated. In a data mesh, tools can be used to enforce global policies such as GDPR enforcement or access management and also local policies where each domain sets its own policies for their data products such as access control or data retention.