What Is a Data Mesh — and Why Enterprise Needs One
A data mesh is a decentralized, domain-oriented architecture for analytical data. Instead of funneling everything into a central data lake owned by one overwhelmed team, each business domain owns, produces, and serves its own data as a product — with shared governance and a self-serve platform underneath.
Domain Ownership
Each business domain owns its analytical data end-to-end — the people closest to the data manage it.
Data as a Product
Treat datasets like APIs — discoverable, documented, versioned, with quality SLAs and clear ownership.
Self-Serve Platform
A shared infrastructure layer that lets any domain team build, deploy, and monitor data products autonomously.
Federated Governance
Global standards for interoperability, security, privacy, and quality — agreed by all domains, enforced as code.
Why Enterprises Are Adopting Data Mesh
Enterprise data has always lived in two worlds: operational systems that run the business in real time, and analytical systems that look backward for insight. Zhamak Dehghani calls this the great divide of data — and the traditional answer of funneling everything into a central warehouse or lake only works at modest scale. As organizations grow in source systems, consumers, and use cases, the central team becomes a bottleneck, not an enabler.
Data mesh addresses these dimensions by inverting the topology — organizing data by domains, not by technology stack. As Fowler and Dehghani describe, the four principles are collectively necessary and sufficient: domain ownership localizes accountability, product thinking ensures quality, a self-serve platform removes friction, and federated governance maintains interoperability without centralized control.
The result is not a new technology — it is a new operating model. One where every team publishes its data as a product with SLAs, documentation, and access controls, and a shared platform makes that publication as easy as deploying a microservice.
The Problem — Central Data Team
- ✕ Central team becomes the bottleneck for every request
- ✕ Engineers lack domain context — data quality degrades
- ✕ “ETL spaghetti” grows faster than the team can maintain it
- ✕ Business decisions wait days or weeks for data
The Solution — Data Mesh
- ✓ Domain teams own, produce, and serve their own data
- ✓ Data is discoverable, documented, and versioned like an API
- ✓ Shared platform with federated governance as code
- ✓ Cross-domain composition without centralized bottlenecks
Meet Orange Co
Orange Co
Technology-Driven Commerce · Global Enterprise
Orange Co is an imaginary multi-million-dollar technology-driven commerce company with thousands of employees globally. The company designs, manufactures, and sells both physical products and digital subscriptions — direct-to-consumer through its own e-commerce platform and through a network of wholesale and retail partners.
Like most enterprises at this scale, Orange Co runs 55 operational applications across 7 business domains — SaaS platforms for CRM and marketing, ERPs for finance and supply chain, real-time streaming systems for inventory and clickstream data, developer tooling for engineering velocity, HRIS platforms for people operations, and customer support systems for post-sale experience. Over time, these systems became data silos: each generating valuable information, but none of it connected.
The result was a familiar pattern: a small central data team trying to serve the entire company, spending most of their time fixing broken pipelines and learning unfamiliar domains. Business questions that should take hours took weeks. Strategic decisions were made on gut instinct instead of data.
That's why Orange Co adopted a data mesh architecture — shifting data ownership to the domain teams who know it best, while keeping a shared platform and federated governance underneath.
How Orange Co Operates
Sell
The sales team manages the full revenue cycle — prospecting via ZoomInfo, engaging through Outreach and Gong, closing in Salesforce, and signing contracts via DocuSign. The SMB segment runs self-serve through HubSpot.
Market
Marketing drives demand through Google Ads, Meta, and LinkedIn campaigns. Segment stitches every customer touchpoint into a unified event stream. Braze and Iterable handle lifecycle messaging for subscribers.
Fulfill
Supply Chain orchestrates end-to-end fulfillment — SAP and Oracle SCM for procurement, Kafka for real-time order and inventory events, Manhattan WMS for warehouse operations, and ShipStation for last-mile delivery.
Build
Product & Engineering builds and operates the platform. Amplitude and Pendo capture product usage. GitHub and Jira track delivery velocity. Datadog and PagerDuty keep systems healthy. Kafka streams real-time clickstream data.
Finance
Finance runs on EBS and Oracle Financials for ledger and GL. Stripe and Zuora handle billing and subscriptions. NetSuite manages revenue schedules. Anaplan powers FP&A planning.
People
HR manages the employee lifecycle through Workday and BambooHR. Greenhouse handles recruiting. Lattice and Culture Amp drive performance reviews and engagement surveys. Deel manages global contractors.
Support
Customer Support ensures every interaction is tracked and resolved. Zendesk manages tickets and SLAs. Intercom handles live chat. Statuspage communicates incidents. Confluence stores runbooks. SurveyMonkey captures CSAT scores.
Domains & Their Applications
Following domain-driven design, we organized the enterprise into seven autonomous domains. Each domain has a dedicated team that knows its data best — and owns the applications that generate it.
Click any domain below to explore its application landscape and the systems that power it.
Finance
8 applicationsThe Finance domain manages all financial operations — revenue recognition, billing, accounts payable, tax compliance, and financial planning. EBS and Oracle Financials handle the general ledger and enterprise accounting. Stripe processes payment transactions from the e-commerce platform. Zuora manages subscription billing — renewals, upgrades, and usage-based add-ons. NetSuite handles revenue schedules and ASC 606 compliance. Coupa manages procurement spend. Avalara automates tax calculations across jurisdictions. Anaplan powers the FP&A team's budgeting and forecasting models.
Connected planning platform. Powers financial modeling, what-if scenarios, and rolling forecasts for the FP&A team.
Automated tax compliance. Calculates sales tax, VAT, and GST across jurisdictions in real time.
Procurement platform. Manages purchase requisitions, vendor selection, and spend management.
Core ERP system. Manages general ledger, accounts payable/receivable, cost centers, and vendor master data.
Financial management for revenue schedules, ASC 606 compliance, and multi-subsidiary accounting.
Enterprise financial management. Handles multi-entity consolidation, intercompany transactions, and statutory reporting.
Payment processing platform. Handles online transactions, subscriptions, invoicing, and payout reconciliation.
Subscription management and billing. Powers recurring revenue models, usage-based pricing, and subscription lifecycle.
Building the Enterprise Mesh
In a mesh, data is not a by-product that gets dumped into a lake — it is a first-class product with an owner, a schema contract, quality SLAs, and documented access patterns. Every data product should be discoverable, addressable, trustworthy, and composable — ensuring consumers can find, understand, and reliably use data across domain boundaries.
At Orange Co, we organize 89 data products into three layers:
Every source application gets a 1:1 data product — a schema-faithful landing zone that preserves the original structure while making it queryable and governed inside the mesh.
Domain teams compose source products into analytical entities — Customer 360, Revenue Ledger, Campaign Performance — combining data across systems into a single trusted view.
Purpose-built for specific audiences and use cases — board-ready dashboards, churn prediction features, demand forecasting APIs, and self-serve analytics for business teams.
Platform Architecture
Orange Co's data mesh runs on a four-stage architecture. Source systems (SaaS, ERPs, streaming, databases) feed into a managed ingestion layer. All 89 data products land in a Cloud Warehouse organized into Source, Business, and Consumer layers — with environment isolation across dev, pre-prod, and production. A self-serve platform provides transformation, orchestration, quality monitoring, and CI/CD so domain teams ship data products independently. Federated governance enforces security, lineage, and data contracts across the entire mesh.
Federated Governance
Governance in a mesh is not a gate — it is a set of computational policies baked into the platform. Domain teams retain autonomy over their data models, but agree to global standards for interoperability, security, and quality. As Fowler puts it: the governance group maintains an equilibrium between centralization and decentralization.
Row & column security per role and environment
Automatic masking of sensitive fields
Time-based archival policies per data class
SOX, GDPR, CCPA — audit trails and consent
Freshness SLAs, uniqueness, referential integrity
Source-to-consumer tracking with impact analysis
Introducing MeshLens
A data mesh at enterprise scale is a living system. Domains evolve, data products multiply, consumers shift, pipelines change, and quality can drift. Static wikis and slide decks quickly fall behind.
MeshLens turns enterprise mesh metadata into an interactive data ecosystem map — connecting domains, applications, data products, lineage, quality, cost, and operational health in one navigable view. Built as a visualization layer on top of mesh metadata, MeshLens helps teams make the ecosystem easier to explore, measure, and communicate — from CTO-level architecture reviews to day-to-day analysis by data product owners, platform teams, and analysts.
Navigate the entire data mesh as an interactive, living map. See how domains connect, which pipelines are healthy, where data flows — and why this architecture matters. An interactive experience that turns complexity into clarity.
A reference schema design showing how to structure a mesh from ERD to materialized views. Use it as a starting point — adapt the data models, naming conventions, and layer boundaries to fit your own enterprise.