Intuit's Data Mesh - Data Mesh Leaning Community meetup 5.13.2021

Tristan Baker - linkedin.com/in/tristanbaker
Suresh Raman - linkedin.com/in/ramansuresh
Allison Bellah (in absentia) - linkedin.com/in/allisonbellah
Data Mesh at Intuit
May 13, 2021

©2021 Intuit Inc. All rights reserved. 2
Arriving at data mesh
Our vision and four part strategy
Now with 25% more parts!
Q&A
Fire away!
Agenda

A brief history of data infrastructure

Today

What “we cannot scale” sounds like from our users
Discovering Data
● Where can I ﬁnd data about a particular thing (customer,
company, etc)?
● Where can I ﬁnd the data sourced from a particular product
or service?
Understanding Data
● Who can approve my access so that I can see samples of the
data?
● What is the schema of the data?
● What is the business meaning and context of the data?
● Is this data related to other concepts? Is it joinable to other
data? What is the meaning of the relationship?
Trusting Data
● What system produces this data and at what latency?
● What other systems use this data?
● What is the quality of this data? Is it ‘clean’?
● Which team supports this data if it breaks?
Publishing Data
● How do I describe my data so that others understand what it
means and how to use it?
● Where do I host my data so that other systems can access it?
● Data systems are complicated, how can I build and operate
my process on top of one?
● What are my operational responsibilities once my
process/data is in production?
● How do I meet my compliance requirements for
processing/storing/publishing data?
● Am I duplicating processing/data that already exists?
Consuming Data
● How is this table/topic partitioned?
● Who can approve my production system to access it?
● Will I get alerted if the schema changes?

The future of data infrastructure
● Data treated as code
● Data service as a facet of a product
● Data responsibility decentralized
● Producers take responsibility for data
● Producers serve consumers
● Data platform provides the ecosystem to
govern and manage the lifecycle of data and
machine learning
The provocation
Data Mesh is born

Our vision and four part
strategy

Enable more Intuit teams
to more easily use and
create data

Four part strategy
• Stewardship
– ensures accountability for a set of defined responsibilities in building and managing their solutions; including
adherence to a set of defined best practices to produce only high quality data.
• Organizing people, code and data
– A systematic approach to organizing the people, code and data which clearly identifies the owners of a business problem and its
solution.
• Self serve products
– A rich suite of self serve products that enable teams to more easily author, deploy, govern and operate their own solutions, aided
by automation and processes that support best practices and high quality as a precondition for deployment.
• Rationalizing data definitions
– A process for rationalizing all critical data definitions at the company so that data concepts like Customer, Product and
Entitlement are unique, re-usable and non-conflicting.

Stewardship goals for next year

Organizing People,
Code, and Data

Raw information about physical systems that describes where the data is stored and where code is
executing. This describes where data is physically located so that it can be accessed.

Basic dependency, ownership and classiﬁcation information provides additional context about physical
data and code locations so that data can be better governed, secured and operated by the owning
teams.

Why organizing people, code and data matters
19
Private vs Public
~50% tables are either
temp/sandbox/staging/test/backu
p tables
- Messes up search & discovery
- Teams consume data not meant
for external use
Data Ownership
~50% tables don’t have clearly
identiﬁed owners
- Erodes Trust
- Copies proliferate
- Operational, Governance risk

Data Processing Capabilities Data Serving Capabilities

Self Serve goals for next year
100% of Top 20 tasks in the Data lifecycle are Self Serve
Infra Provisioning
● Transactional Persistence
● Compute for stream, batch
processing
● Monitor, Debug Infra
● Cost
Data Authoring
● Events, Schemas
● Ingestion
● Transformations
● Entities
● ML Features
● Data Quality,
Observability
● Orchestration
Data Governance
● Access Management
● Key management
● Compliance Controls &
Audit
● Privacy

Rationalizing Data
Deﬁnitions

Clean entity information with formally deﬁned meaning and relationships enables better data understanding. This
is the purpose of entity deﬁnitions. They ensure that data is clean, organized, connected, discoverable and
documented in a formal way.

When you bring it all together, you get Intuit’s Data Mesh

Capturing meaning,
relationship, ownership, and
system dependencies builds a
full, rich picture for everyone.
No tribal knowledge needed!
In this example, the clean
information describes entities
OII Account and Intuit Product
and the Entitled To relationship
between them.
The basic information describes
how the data for these entities
are sourced from the Identity
Universal Service and the
Entitlement Reference Service.
The raw information describes
which Event Bus topic and Data
Lake table the data for these
entities can be found in.

Q&A
Tristan Baker - linkedin.com/in/tristanbaker
Suresh Raman - linkedin.com/in/ramansuresh
Allison Bellah (in absentia) - linkedin.com/in/allisonbellah

32

Intuit's Data Mesh - Data Mesh Leaning Community meetup 5.13.2021

More Related Content

Intuit's Data Mesh - Data Mesh Leaning Community meetup 5.13.2021