Microsoft Fabric

Unified Analytics for

Data-Driven Innovation
A proof-of-concept framework for data leaders considering Fabric
Contents
Chapter 1 03
What is Microsoft Fabric?

Chapter 2 05
Your data analytics needs

Chapter 3 10
Fabric success stories

Chapter 4 12
Building a proof of concept for Fabric

Take the next step 17

 Microsoft Fabric: Unified Analytics for a Data-Driven Innovation Landscape 3

What is Microsoft Fabric?

Your company’s data contains untold value – and
the tools you use to manage and analyse it are
key to unlocking its potential.
Why Fabric?
Today’s data and IT leaders want data solutions that
simplify the process of extracting value from data, Despite collecting vast amounts of data
allowing their teams to work quickly, confidently from apps, services and IoT devices,
and securely. With the right tools, they know that many businesses struggle to turn it into
they can help teams deliver better insights and actionable insights. Data analysis tools
maximise the value of their data without unnecessary often require specialist expertise, creating
complexity. bottlenecks as users must rely on experts
to deliver insights.

That’s where Microsoft Fabric Moreover, data silos across lines of

business – HR, marketing, sales, finance,
comes in. etc. – further complicate decision-making,
leading to inefficiencies and missed
Fabric is a Software-as-a-Solution (SaaS) platform opportunities.
that delivers an integrated and simplified experience
on an open and governed foundation. It offers As an organisation grows, managing
intuitive tools for streamlining analytics workflows diverse data tools becomes increasingly
and serves as a single source of truth for data, complex and costly, diverting resources
making it easier for teams to collaborate and deliver from core priorities. This complexity calls
more impactful insights. for a streamlined approach that simplifies
data management and closes the gap
Use this eBook to understand Fabric’s core between teams and the insights that they
capabilities, examine your unique analytics needs need to thrive.
and create a proof of concept for using Fabric to
help your team get more valuable insights, faster. Adopting a unified data platform like
Fabric is essential to overcoming the
data challenges hindering AI innovation
and blocking access to insights. With
a data foundation built on Fabric, your
organisation can better unlock actionable
insights, accelerating critical decision-
making and driving sustained growth.
 Microsoft Fabric: Unified Analytics for a Data-Driven Innovation Landscape 4

What's hindering
How Fabric helps
access to insights?

Data is dispersed across multiple platforms, Unify your data in Onelake. As Fabric’s open
creating silos and making it difficult to access and governed data lake foundation, OneLake
real-time information. provides a single source of truth so everyone
is working with the same data. With multiple
integration options, your teams can bring
their data to OneLake without duplication
or movement.

Teams lack sufficient data literacy and Empower your teams with democratised
confidence to derive meaningful insights from access. Providing secure access to the right data
their data. and analytics tools at the right time empowers
your teams to gain insights faster, leading
to better data-driven decisions.

Managing a complex web of different data tools Integrate your solutions on an open and
and suppliers increases operational costs and governed foundation. Fabric unifies all your
security risks while complicating governance. data, no matter where it’s stored, giving your
data stewards full control to manage it from
start to finish, while addressing the long-standing
security and compliance challenges that your
teams have faced for years.
 Microsoft Fabric: Unified Analytics for a Data-Driven Innovation Landscape 5

Your data analytics needs

Simplified data
Every organisation has unique data OneLake is a unified data storage solution within
challenges, but they tend to fall into Microsoft Fabric that acts as a central hub where
five core categories: data can be stored, managed and accessed
effortlessly.
1. Simplifying data-handling

2. Preparing data for AI There are various ways to bring data into Fabric,
including traditional ETL pipelines and mirroring,
3. Providing specific experiences for meaning some methods still involve replicating
different roles data. However, using shortcuts can help reduce
data duplication. OneLake supports managed and
4. Integrating seamlessly with existing tools
unmanaged data storage so users can connect to
5. Maintaining comprehensive governance various data sources without replicating the data.
and security Through symbolic links – or shortcuts—OneLake
enables access to data from different cloud providers
within the same environment, removing the need to
move or duplicate data.

This set-up helps reduce data silos and allows a

Whether you’re dealing with data silos, inefficiencies single copy of data to be used across multiple
or complex workflows, addressing these needs is analytics tools and workflows. For example, data
critical to maximising the value of your data and engineers can create real-time analytics dashboards
getting the best AI outcomes. Fabric provides a without duplicating data, and data scientists can
holistic solution designed to tackle these challenges train machine-learning models using the same data
head-on, equipping teams with the tools they need set without creating multiple copies. This approach
to work faster and smarter. saves on storage costs while also improving data
consistency and integrity across the organisation.

Find out how to create a lakehouse in Fabric.

 Microsoft Fabric: Unified Analytics for a Data-Driven Innovation Landscape 6

A range of analytics
experiences
Data engineering
Fabric offers a comprehensive suite of analytics
experiences tailored to various data use cases, use cases
ensuring that users can seamlessly manage, analyse
and visualise their data. Each experience is designed Speed up Spark applications
for specific roles and enhances productivity and Fabric’s native execution engine – included
insights across the entire data lifecycle. at no additional cost – allows data engineers
to use the familiar Spark ecosystem while
benefiting from faster deployment of

Data engineering: Building Spark clusters. This results in enhanced

performance, reducing the time to market
efficient data workflows for critical applications.

Migrate to Microsoft Fabric

Data engineers play a critical role in designing, Many organisations are migrating their
building and maintaining the infrastructure that existing Synapse Spark applications and
enables efficient data collection, storage and SQL Server or Synapse Dedicated Pool data
processing. Fabric simplifies these tasks, allowing warehouses to Fabric, taking advantage of
engineers to focus on getting value from their data. faster deployment and pricing flexibility.
Fabric simplifies the migration of T-SQL
Data Factory code, making it a robust option for teams
Use over 150 connectors to design, schedule and seeking a more streamlined, scalable
orchestrate data pipelines, enabling seamless Extract, solution.
Transform and Load (ETL) processes for efficient data
integration.

Data Engineering
Develop large-scale data solutions using Apache
Spark and other big data technologies to handle
complex data workloads.

Data Warehouse
Create and manage data warehouses that are
optimised for query performance and data
integration.
 Microsoft Fabric: Unified Analytics for a Data-Driven Innovation Landscape 7

Data warehousing:
Centralising and
optimising data
Data warehousing is key for organisations looking
to centralise their data for analysis and reporting.
Fabric enhances this process by enabling efficient
migration and management.

Data Warehouse
Query and analyse large data sets – such as
sales data – to identify trends and patterns.

Real-Time Intelligence
Gain access to real-time insights through
up-to-the-minute data analysis.

Real-time data analysis: Real-time analytics

Immediate insights for use cases
critical decisions Point-of-sale data streaming
Retail businesses can use Fabric to stream
For businesses requiring real-time data insights,
real-time point-of-sale (POS) data from
Fabric offers tools that enable instant access to
multiple stores into a centralised location.
streaming data, empowering faster and more
This capability – powered by Event Streams,
informed decision-making.
Kusto DBs and Data Activator – allows
retailers to generate real-time reports,
Real-Time Intelligence
enabling quicker, more strategic decisions.
Build real-time solutions that can ingest and analyse
streaming data. Time series analysis of manufacturing
historian data
Data Activator
When dealing with large volumes of device
Automate alerts and triggers for in-motion data
telemetry, much of it is stored and often
based on specific criteria.
aggregated in a manufacturing historian.
Fabric allows you to stream this historian
data to OneLake, enabling advanced analytics
like predictive maintenance, simulations and
cross-site comparisons.
 Microsoft Fabric: Unified Analytics for a Data-Driven Innovation Landscape 8

Data governance: Ensuring Machine learning: scaling and

compliance and security automating AI solutions
Fabric’s integration with Microsoft Purview Fabric empowers data scientists and business users alike
Data Governance provides comprehensive data to take advantage of machine learning and AI, integrating
governance, allowing organisations to classify, tools that streamline model development, training and
protect and manage their data with ease. You can deployment.
classify Fabric items, configure data-loss prevention
and monitor user actions to identify and mitigate Data Science
potential security risks. Develop and deploy machine-learning models within
an integrated environment.

Azure Machine Learning

Use advanced machine-learning capabilities to experiment
with algorithms and optimise models.

Governance use cases

Protect sensitive data
Using Microsoft Purview’s Information
Protection and Data Loss Prevention policies, Data Science use cases
organisations can automatically detect and
protect sensitive data in Fabric, ensuring Transform large-scale unstructured data
compliance with regulatory requirements. Organisations with large volumes of unstructured
These tools prevent unauthorised access or semi-structured text, such as customer service
and restrict the export of sensitive data, transcripts or legal documents, can use Fabric’s
safeguarding the organisation from potential integration with Azure OpenAI to derive insights.
breaches. Scalable AI functions enable users to process these
data types, extracting complex information such as
Learn more about governance in Fabric. tables and structured elements from large PDFs or
audio recordings.

Deliver customised generative AI experiences

for your data
An AI skill in Fabric allows your users to talk to
your data in OneLake from your own applications.
This ‘data agent’ is configured as a conversational
Q&A chatbot on your data domain. By providing
instructions and examples to guide the AI, creators
can ensure that the ‘data agent’ understands the
organisation’s data context and provides a reliable
data-driven response in return.
 Microsoft Fabric: Unified Analytics for a Data-Driven Innovation Landscape 9

Reporting and visualisation: Integrations

Driving actionable insights
Your data platform needs to work with
Non-technical business users often require easy the tools your teams already use. Fabric
access to data insights. Fabric ensures that data can integrates deeply with Microsoft Office,
be visualised and shared seamlessly regardless of Teams and other commonly used apps to
technical expertise – or lack of it. ensure that insights are easily accessible
across your existing workflows.
Power BI
Create interactive reports and Whether it’s through Excel, Word or Power
dashboards, visualise data and BI, you can analyse, share and collaborate
share insights. on data directly within the applications
your team is already comfortable using,
Copilot for Power BI simplifying the data process and enabling
Use natural language prompts to quicker decision-making.
generate reports and insights, making
data accessible to non-technical users.
Governance and
security
Data stewardship and security are critical
as data environments become increasingly
Reporting and complex. Fabric simplifies this with built-
visualisation use case in security and governance features,
minimising the effort needed to safeguard
Cost savings through tool consolidation your entire analytics platform and reducing
By consolidating multiple data visualisation the risk of data breaches and compliance
and management tools into Fabric, issues.
organisations can save on costs and reduce
technical debt. Fabric’s Direct Lake mode OneLake provides governance tools like
allows users to mirror data sources, reducing data lineage, data protection, certification
the strain on existing data warehouses while and catalogue integration to help
maintaining access to real-time analytics. streamline data management. These tools
offer full visibility over your entire analytics
environment, from data sources to user
insights. Meanwhile, your tenant admins
can easily maintain control over how your
data is accessed and shared.
 Microsoft Fabric: Unified Analytics for a Data-Driven Innovation Landscape 10

Fabric success stories

Government

Gamle Oslo enables unified

Fabric makes analytics accessible to people
data visibility without a technical background. It makes
this data accessible and means that we
The Norwegian government is focused on creating have a stepping stone to adjust, redevelop
the best possible society for its citizens. To better and improve our services for residents.”
manage its data, Gamle Oslo sought to centralise
Aleksander Lorentzen,
and gain insights from it. The district began using
Head of Digitisation,
Fabric as a unified platform, and after positive results
District of Gamle Oslo
from using it with a few data sets, they now plan to
expand it across the entire district.

Agriculture

The diversity of tools in Microsoft Fabric

Alltech centralises data and
allows us to choose the right tool for the streamlines travel spend
job. OneLake allows data consolidation
from a variety of data sources. Alltech
analysis
received top-notch technical and best
Alltech, a global leader in sustainable agriculture,
practice support from both Microsoft
needed to bring together data from its various
and Sonata.”
acquisitions to track travel expenses better and
Connie Thompson support its sustainability goals. Partnering with
Global Director of International Sonata, an early adopter of Fabric, Alltech integrated
Technology, Data, Analytics data from multiple systems into one place,
and Reporting centralising its data and gaining valuable insights.
Alltech The pilot project succeeded in managing travel
costs and has set the stage for Alltech to continue
modernising its data strategy and empowering users
with self-service analytics.
 Microsoft Fabric: Unified Analytics for a Data-Driven Innovation Landscape 11

Retail

Chalhoub Group’s People We took full advantage of the six-month

Analytics team accelerates trial during the Fabric preview. We had
anticipated some issues arising while
reporting Microsoft was still upgrading the platform.
With the evolution of Fabric and incredible
Chalhoub Group, a luxury retailer based in the
support from the Microsoft technical
United Arab Emirates, wanted to streamline its HR
support teams [and] Fabric specialist[s], all
approval process and eliminate the need to gather
of those challenges disappeared, and we
data from multiple platforms. After testing Fabric,
were set up for success.”
the group modernised its HR data analytics by
consolidating all data into one platform, improving Mark Hourany
agility, enhancing analytics and speeding up Director, People Analytics,
processes. Chalhoub Group

Professional Services

The innovative ability of Microsoft Fabric ITOCHU Corporation

to import data virtually from many launches a data integration
sources will establish it as a key platform
for supporting diverse applications of platform to accelerate AI
generative AI. We want to leverage this
foundation to promote generative AI and With more than 70 ideas for using generative AI,
other advanced technologies over the long ITOCHU wanted to develop a data integration
term.” platform to bring these ideas to fruition. The
company looked to Microsoft Fabric to help
Yusuke Tsujii aggregate and cleanse data to make it more
ICT Division task force leader suitable for AI use.
ITOCHU Corporation
 Microsoft Fabric: Unified Analytics for a Data-Driven Innovation Landscape 12

Building a proof of concept

for Fabric
A proof of concept is a trial used to demonstrate
the feasibility of a solution in a real-world scenario.
2. Set a timeline
It’s essential to determine if a platform meets your
A proof of concept should be a time-bound
organisation’s data management and analytics
exercise with specific, measurable goals. Based
needs. A successful experimental period will show
on best practices, it’s recommended that you
the potential for production use by validating
timebox your trial to about two weeks. This will
technical solutions like system integrations or
provide enough time for you to see meaningful
specific configurations.
results while limiting complexity.

Follow this basic framework to see how your

Suggested timeline:
organisation could realistically create and
execute a proof of concept for Fabric. • Data loading: three days or less

• Querying: five days or less

1. Identify sponsors and • Additional tests: two days or less

potential blockers When managing your timeline, making realistic

estimates for task completion is important. The
To ensure that your trial proceeds smoothly, scope and timeline can vary based on factors
it’s essential to secure the necessary support like data set size, the complexity of database
and identify any roadblocks early on: objects and the number of interfaces being
tested. If the estimated time exceeds four weeks,
• Identify any organisational restrictions or
consider narrowing the focus to prioritise
guidelines for moving data to the cloud.
high-impact goals.
• Secure executive and business sponsorship
for the proof of concept.

• Ensure that Fabric is suitable for your workload.

 Microsoft Fabric: Unified Analytics for a Data-Driven Innovation Landscape 13

3. Define the architecture 4. Identify current pain points

and parameters and set goals
Design a high-level architecture for the Your experimental period should address specific
proof of concept that includes relevant data pain points or gaps in your current system.
sources, consumers and any required AI or
machine-learning components:

• Decide which components will be part
Consider the following questions:
of the trial and exclude those that aren’t
immediately necessary. What gaps do you expect Fabric to fill?
• Identify existing resources (like Azure What new business needs are emerging?
services) that will support the proof of
What service-level agreements (SLAs)
concept.
must be met?
• Choose an appropriate region and
What types of workloads (e.g. ETL,
subscription for non-production use.
analytics, reporting) will the proof
• Ensure that your network infrastructure of concept focus on?
can handle the proof-of-concept workload
without affecting production systems.

When considering migration scenarios from

After you’ve identified your pain points, the
a legacy data warehouse, evaluating your
next step is to define the goals of your trial
approach is important. Are you aiming to
clearly:
make minimal changes to your existing ETL
processes, or are you open to making significant • E
 nsure that query performance meets new
improvements during the migration? Service Level Agreements (SLAs).

• Assess system responsiveness for end users.

Alternatively, you might be building an entirely
• E
 valuate if current ETL processes work well with
new environment, known as a greenfield project,
Fabric.
from scratch. Each scenario requires careful
planning to ensure a smooth transition. • E
 nsure robust governance and security
compliance.
 Microsoft Fabric: Unified Analytics for a Data-Driven Innovation Landscape 14

5. Create a test plan

Here are some examples of Fabric trial
Based on your established goals, design goals and how they can be tested:
specific tests to validate the performance and
capabilities of Fabric. Each test should align with Goal: Ensure that query performance
a particular goal, with clear expectations for the meets new SLAs.
outputs. Plan for specific queries, reports, ETL
 est complex queries under varying loads
T
jobs and other processes to be executed on a
to verify that they can handle the expected
defined data set. It’s also essential to involve all
data volume and complexity within the
stakeholders in agreeing on the test plan and
required timeframes.
success criteria to ensure alignment and clarity
throughout the process.
Goal: Assess system responsiveness
for end users.

Measure latency and throughput of queries

6. Identify and validate the executed by users to ensure a smooth,

proof-of-concept data set efficient experience.

Goal: Evaluate if current ETL processes

From the tests you’ve planned, identify the data
work well with Microsoft Fabric.
required in Fabric to execute them effectively.
Analyse ETL workflows for compatibility and
Take some time to review this data set with opportunities for improvement, focusing
the following considerations: on transformation logic and data loading
processes.
• Ensure that the data set accurately represents
the future processing on Fabric in terms of Goal: Ensure robust governance and
content, complexity and scale. security compliance.
• Avoid using a dataset smaller than 1 TB, as it Test data access controls, encryption and
may not provide representative performance, compliance with regulatory standards to
and avoid overly large data sets, unless ensure that the platform meets organisational
the proof of concept is specifically testing security policies and industry regulations.
processing capacity.

• Determine the appropriate distribution

pattern and indexing option for each table.
If there are any uncertainties regarding
distribution, indexing or partitioning, include
tests in your proof of concept to address
those questions.
 Microsoft Fabric: Unified Analytics for a Data-Driven Innovation Landscape 15

7. Assemble your team 8. Run key workflows

Ensuring that the right people are involved in Test the essential workflows that are part
the experimental period is critical to its success, of your trial:
as each role uniquely guides the project from
• D
 ata engineers should test data integration
planning to execution.
methods and transformations – for example,
as in medallion architectures.
A diverse team brings together the necessary
skills and expertise to tackle both technical • Data scientists can train machine-learning
and business challenges: models using the same data set to ensure
team efficiency.
• A project manager to lead the effort.
• Business analysts can use generated
• A business representative to oversee
reports and insights, testing ease of use,
requirements and results.
customisation options and the ability to drill
• A data expert familiar with sourcing the data down into their data for deeper analysis.
for the proof of concept.
• IT and security teams should test role-based
• A Fabric specialist and potentially an expert access controls and data masking features
advisor to optimise tests are needed. to ensure that data access permissions are
• Other specialists, such as network or Azure correctly configured.
administrators, for specific components of • End users should interact with dashboards
the trial. and reports to test their usability and
accessibility, taking note of the ease of
Since you are evaluating a new platform, navigation and clarity of information.
consider engaging a Microsoft partner or
consultant with expertise in Fabric to assist
with the proof of concept.
 Microsoft Fabric: Unified Analytics for a Data-Driven Innovation Landscape 16

9. Measure results
To measure the results of your trial, track key
metrics such as data processing speed, capacity
levels and workflow efficiency. These metrics
will provide valuable insights into how well
Fabric performs in your environment. Focus
on identifying improvements in operational
efficiency, cost savings and data consistency –
especially the advantages of unifying all your
disparate data to one central data lake.

By assessing these factors, you’ll be able

to determine how effectively the platform
streamlines processes and reduces the
complexity and costs associated with
data management.

10. Evaluate results

Once your trial is complete, review the
outcomes against the initial goals:

• Have the goals been met?

• Were the expected outputs achieved?

• Identify areas where further testing or

evaluation may be required and document
any new questions raised during the
experimental period.

This structured approach to a Fabric proof

of concept will ensure a clear validation of
the platform’s capabilities in addressing your
organisation’s specific challenges and goals.
 Take the next step
Now that you understand what Microsoft Fabric is
and how it helps organisations get more value from
their data, it’s time to take the next step – testing
your Fabric proof of concept.

Whether you’re facing issues like data silos,

bottlenecks in decision-making or high operational
costs due to complex data systems, Fabric has the
tools to overcome these hurdles.

• Establish a single source of truth for your data

to break down silos.

• Democratise data access and empower teams

to make data-driven decisions.

• Connect
 and govern your data foundation
to reduce complexity and ensure security and
compliance.

Ready to unlock your data’s potential?

Start a trial run of Microsoft Fabric today and see how it can transform your organisation’s
decision-making process.

Sign up for a free trial

© 2024 Microsoft Corporation. All rights reserved. This document is provided ’as is’. Information and views expressed in this document, including
URL and other Internet website references, may change without notice. You bear the risk of using it. This document does not provide you with any
legal rights to any intellectual property in any Microsoft product. You may copy and use this document for your internal, reference purposes.