Swisscom is taking one bold step after another to become a data-driven company. The approach is always business-first: how to find data&AI-driven solutions to enable the business to make the best decisions to offer a great experience to our customers. Our journey with Data Mesh is no different. Together with the business, we looked at the current challenges of quickly transforming data into information and insights while having a crucial regard for data management and governance. I invite you to this session to go through our transformation from data to data products, how to foster co-creation between data producers and data consumers, and what it takes to create the right balance between central governance and decentralizing the accountability for its implementation.
Creating a Product through DevOps: The Story of APPUiO Cloud
BATbern52 Swisscom's Journey into Data Mesh
1. Swisscom's Journey into Data
Mesh
Mirela Navodaru, Enterprise and SolutionArchitect Data, Analytics and AI, Swisscom
Berner Architekten Treffen, 08.03.2024
3. Fast Insights
Together,
We Make
Data Work
Data Driven
Culture
Value Driver
Data
Skills
SC Data
Community
Data-Driven
Experience
Data-Driven
Operational Excellence
Data-Driven
Innovation & Growth
Data Mgmt.
Platform
Data centric Roles &
Responsibilities
Data Security,
Compliance & Quality
Data
Literacy
One Data
Platform
Data
Products
Self-
Service
The Swisscom Data and Analytics Strategy
Generating Business Value from Data at Scale
Target
Picture
Best Data
Management
4. Mirela
Navodaru,
Swisscom's
Journey
into
Data
Mesh,BAT52,
08.03.2024,
C1
Public
4
From the Starting Point to the Data Democratization
Data
as a Product
Federated Data Governance
Self-serve and domain-agnostic
Data Infrastructure as a Platform
Data Management Central Suite
Centralized Data, Analytics & AI
Delivery Organization
Data
as an Organizational Asset
Data managed
in Silos & Redundant mode
in an Overcomplex Landscape
Central Governance vs Delivery
Data, Analytics & AI
+
Decentralized Data Products Teams
?
Data Mesh enables Data Democratization
Data
Mesh
Pillars
5. Mirela
Navodaru,
Swisscom's
Journey
into
Data
Mesh,BAT52,
08.03.2024,
C1
Public
5
Data Products – What it means for Swisscom
Value Proposition – a Must
Reusability
Interoperability
Use the Manifest to
document Data Products
Use the Data Contract(s)
Conditions for Sharing Data
Enable Data Protection &
Security by Design
The Output - Data Sets (low granularity data, aggregated data, transformed data, insights from
ML applied techniques, etc.) with a Value Proposition.
Principles
7. Mirela
Navodaru,
Swisscom's
Journey
into
Data
Mesh,BAT52,
08.03.2024,
C1
Public
7
Data Products – Deep Dive into types
Data Sources Business cases
Customer-aligned
Data Products
Conformed
Data Products
Source-aligned
Data products
Characteristics:
• Lowest level of granularity
• Structure and Model from the
Data Sources
• Data cleaned according with
the data quality rules
• Target state OP –
(transactional system) cross-
functional teams with
business, data & tech skills
Characteristics:
• Conformed with enterprise
logical data model (SDM)
• Greenfield approach,
governance built-in
(TMForum);
• Target state OP –cross-
functional teams with
business, data & tech skills
(Organized by Domains)
Characteristics:
• Dimensional data model
• Created by merging two or
more Conformed DPs
• Already in place, to be
migrated to the new
platform (almost 1:1)
- Target state OP - cross
teams with business, data &
tech skills
9. Mirela
Navodaru,
Swisscom's
Journey
into
Data
Mesh,BAT52,
08.03.2024,
C1
Public
9
Swisscom Data Management – Metadata and Policies
A Central Meta Data Mgt
System provides information
on data properties, business
context and lineage
Meta Data is automatically discovered from data source
systems or obtained from decentral Meta Data Systems
The Swisscom Data Catalogue
enables transparency and fast
retrival of data and insights
Business, Quality, Security
and Compliance Data
Requirements are centrally
managed as rules
Data Product Cockpits
monitor security, compliance
and quality of the data
Data Governance processes can
proactively be triggered and partly be
automated based on data monitoring and
data catalogue information
Clearly defined data roles and
responsibilities enable efficient data
management and trust in data
10. Mirela
Navodaru,
Swisscom's
Journey
into
Data
Mesh,BAT52,
08.03.2024,
C1
Public
10
Data Products – Metadata to Document and Share
Manifest
Documentation of Data Products
To be created together with the Data Product and maintained during
its lifecycle
Descriptive Elements
Compliance Aspects
Reference Metadata
Data Contract
Agreement between Data Producers and Data Consumers
It describe the Service's Interfaces and the Conditions to share the
Data Products
Exposure Patterns
Commitment Elements
Data Quality Checks
Ownership & Accountability
Measures & Rating
11. Federated Data Governance – Shared Responsibility
Mirela
Navodaru,
Swisscom's
Journey
into
Data
Mesh,BAT52,
08.03.2024,
C1
Public
11
Governance
Board
Central Governance is
easy to implement
(Built-in/by Design)
Assessment of Risks and
Approval of Exceptions
Feedback Loop between
Governance Orgs and
Delivery Teams
Global Intradomains
Governance
Principles and Guardrails
Rules, Policies and
Regulations for
Compliance Purpose
Architectural References,
Standards and Patterns
Team with People from
both Governance Orgs
and Delivery Teams
Local Interdomains
Governance
Ownership and
Accountability for Data
Products Lifecycle
Ownership of Manifests
and Data Contracts
Intradomain Data
Quality and Operations
Conceptual and Logical
Data Model (SDM)
Accountable for the
Governance
Implementation
Global
Interoperability
needs Global
Governance
Autonomy,
Agility and
Flexibility of the
Domains
12. Next steps in our Journey on Data Mesh
✓ Pilots for e2e Data Products Value Chain
✓ Governance Body for Federated Data Governance
✓ Target Operating Model
Mirela
Navodaru,
Swisscom's
Journey
into
Data
Mesh,BAT52,
08.03.2024,
C1
Public
12
13. Key Learnings from our Journey on Data Mesh
✓ Decentralize to bring more Power into Producing and Consuming Data
✓ No Value, no Party
✓ Collaboration over Data and Metadata is the glue for the Data Producers and Consumers
✓ Design the Governance Implementation to be a real Enabler to the Data Democratization
Mirela
Navodaru,
Swisscom's
Journey
into
Data
Mesh,BAT52,
08.03.2024,
C1
Public
13