Enter your search text in the box above

Select a result to preview

Building modern data architectures with Microsoft Fabric

#fabcon2026 #session #learning #microsoft #fabcon2026-session

Description

Learn how Microsoft Fabric can be leveraged with the architectures: Data fabric, data lakehouse, and data mesh. Get a tour of each architecture and explore their pros and cons. By the end of the session, you’ll have a clear understanding of how to use Microsoft Fabric to build and optimize these diverse data architectures.

Key Takeaways

My Notes

Action Items

Slides

FABCON 2026
The Hidden Strategic Leverage in
Modern Data & AI Architectures
Why open architectures havenʼt translated into real
choice — and how to fix it
March 20, 2026
The Bad Days Of Closed Platforms Are Behind Us.
The model most vendors sold to organizational leaders
Platform = Data + Compute + Governance + Workloads
●
●
●
●
●
●
●
●
●
One platform “ownsˮ everything
Proprietary data and storage formats
Charges to access your own data
Cost of compute completely obscured
Governance is hard-coded into platform
All key workloads provided by platform
Choice is made once, then locked in
Impossible to negotiate
Impossible to experiment
The closed platform model is no longer an acceptable reality
How Did We Break The Closed Platform Model?
Four structural changes have already happened
Open data formats
Governance operates above compute
Multiple viable compute layers
Key workloads now run on any compute
The $200B Paradox: If Open Architectures Are Here,
Then Why Don't Customers Feel Free?
+
Open formats & data
+
Multiple clouds

Portable governance
Still no leverage
The Symptoms Everyone Recognizes.
If this sounds familiar, you’re not alone
Compute costs feel uncontrollable
Consumption is policed
Negotiations lack leverage
AI treated as “specialˮ
Platform decisions feel irreversible
Letʼs dig into it.
News Flash: The Modern “Databaseˮ Is No Longer a Storage System.
It Is a Compute System.
Compute is 90% of the monthly bill
Everything Else
And here’s why
Compute has always been the most expensive part of the
data stack. You just couldnʼt see it clearly before.
Compute is where all the work happens:
10%
Ingest
Transform
Query
AI & agentic execution
As workloads become more powerful, compute becomes:
90%
More specialized
More dynamic
More expensive
Compute
Storage, by contrast, has been largely decoupled:
Open formats
Object storage
Cheap, scalable, commoditized
News Flash: Compute Lock-in Exists Today,
Driven Not By Architecture But By Vendor Friction.
YESTERDAY’S PROBLEM
TODAY’S PROBLEM
Architecture
Vendor Friction
Compute lock-in
Data lock-in
Governance lock-in
Business workloads lock-in
Examples of vendor friction to compute choice [Not architectural]
“Defaultˮ compute engines
Proprietary dialects
Unsupported infra, clouds
Opaque cost visibility
Coupled governance & compute
“Native-onlyˮ AI
Reluctant cost controls
Preferred data formats
Closed orchestration
The False Trade-Off We Must Keep Rejecting.
Central Governance
Security • Compliance • Control • Standards
Visibility and Cost Control
Usage• Attribution • Forecasting
Ease of Management
Training • Provisioning • DevOps
Innovation and AI
Speed • Risk • ROI

Giving up Compute Choice
One vendor • One platform • No choice
The False Trade-Off We Must Keep Rejecting.
Central Governance
Security • Compliance • Control • Standards
Visibility and Cost Control
In open architectures, you donʼt have to
Usage• Attribution • Forecasting
Giving up Compute Choice
give up critical=choice.
One vendor • One platform • No choice
Ease of Management
Training • Provisioning • DevOps
Innovation and AI
Speed • Experimentation • ROI
Not Using Compute Choice Is Strategically Expensive.
We Are Architecturally Free, But Not Acting Like It
Not Using the Freedom Is Expensive
●
Analytics & AI becomes budget-constrained
●
No negotiating leverage with vendors
●
Platform decisions feel irreversible
●
Sovereign, private/hybrid options are lost
Not Using Compute Choice Is Strategically Expensive.
Vendors are pushing Platform = Coupled Compute + Governance + AI workloads
The deadliest danger is recreating the
“Closed Platformˮ by tightly
coupling compute with data
governance and AI workloads
Whatʼs Missing Is Not Another Lakehouse Platform —
Itʼs an Independent Compute Layer!
A standards-based, plug-and-play compute engine that
Works with your existing:
Hosting/cloud layer
Data and governance layer
Business Workloads layer
Requires no re-platforming.
No migration. No lock-in
Delivers measurable impact
Native execution for BI and
AI/agentic workloads
with no platform or format bias
THE RESULT
50% reduction in compute cost, without sacrificing performance, governance, visibility, or control.
The unlock: New architectural principles that have matured over the past decade
What This Looks Like in Practice.
With new architecture, you can deliver the same performance with 50% of the cost
What This Looks Like in Practice.
With Compute Lock-in
With Compute Choice
Minimal savings without performance trade-off
50% compute savings with no perf penalty
All workloads must move to realize savings.
Usually a 3 year re-platforming
Only high-priority workloads can move,
realizing value in a quarter
Migration of data and business workloads
vs
No migration of data or business workloads
Limited agentic support typically only on
favored platform/ format
Agentic workloads across platforms/formats
Execution typically limited to public cloud only
Run across private, hybrid and sovereign
clouds
Open architecture creates possibilities.
Compute choice delivers strategic leverage.
Are you architecturally open or economically free?
KEY TAKEAWAYS
What You Can Do…
Follow the money.
Tally the friction.
Demand a plan.
See that compute is
90% of your platform bill.
Identify what prevents
real choice in compute.
Insist on a strategy for
independent compute.
E6DATA
Thank You!
Turn architectural freedom into economic freedom.
Visit us at Booth #603.