← All Blogs
Data Engineering · Cloud Migration

RDBMS/Data Warehouse to Cloud Migration: Lessons from the Field

A successful migration is a business transformation initiative disguised as a technology project. The organizations that succeed focus on architecture, governance, and business outcomes. The ones that fail focus only on moving tables.

Mudit Kumar
Mudit Kumar · 9 min read

Before You Migrate: Ask the Right Question

Over the last decade, I've worked on multiple data modernization initiatives involving Oracle, SAP HANA, PostgreSQL, Vertica, traditional data warehouses, Hadoop ecosystems, and modern cloud platforms. One question consistently comes up:

"How do we migrate to a modern cloud platform without creating another expensive legacy platform in the cloud?"

Unfortunately, many organizations treat migration as a technology replacement exercise. It is not.

The wrong question:

"How do we move Oracle/HANA/Teradata to Snowflake?"

The right question:

"What should our target-state data platform look like in 3-5 years?"

Migration should never be driven by:

Migration should be driven by:

Three Migration Approaches

THREE MIGRATION STRATEGIES — EFFORT vs VALUE LIFT & SHIFT Fast · Low Effort · Low Value Oracle / SAP HANA Schema Conversion Snowflake (same design) ✓ Fastest execution ✓ Low disruption ✗ Legacy design persists ✗ Technical debt remains ✗ No AI readiness Temporary Landing Only REPLATFORM Medium Effort · Good Value Legacy Warehouse Cloud Landing Zone Snowflake + ELT Modern BI + Analytics ✓ Better cloud utilization ✓ ELT over ETL ✓ Reduced maintenance Sweet Spot for Enterprises MODERNIZATION ★ High Effort · Highest Value Business Domains Data Products Lakehouse Architecture Analytics + AI + Agents ✓ Future-proof ✓ Eliminates tech debt ✓ AI-ready from day one ★ Recommended Low Value, Fast High Value, Slower
Fig 1: Three migration strategies — from quick lift-and-shift to full platform modernization
Approach 1

Lift & Shift

Move schema and jobs with minimal changes. Fastest path but carries legacy design into cloud. Use only as a temporary landing zone.

Approach 2

Replatform

Keep business logic, modernize infrastructure. ETL→ELT, on-prem→Airflow, legacy BI→Power BI/Tableau. Sweet spot for most enterprises.

Architect Insight: Migration is the best opportunity to eliminate years of accumulated technical debt. Do not waste it by simply moving tables to a more expensive location.

Decision Matrix

FactorLift & ShiftReplatformModernization
CostLowMediumHigh (upfront)
TimeFast (weeks)Medium (months)Longer (quarters)
RiskMediumMediumHigher (managed)
Long-Term ValueLowMediumVery High
AI ReadinessNonePartialFull
GovernanceInherited (poor)ImprovedDesigned-in
Cost OptimizationMarginalGoodExcellent

Key Areas Often Ignored During Migration

1. Data Quality

Common assumption: "Bad data will somehow become good data in the cloud." It won't. A modern cloud warehouse is not a data quality solution. Profile your data, identify duplicates, resolve master data issues, and define ownership before migration.

2. Metadata and Lineage

Most migration failures happen because nobody knows where data originated, who owns it, or which report consumes it. If lineage is not documented before migration — expect surprises post go-live.

3. Security and Access Model

Do not migrate permissions blindly. Redesign from scratch:

4. Cost Management

Myth: "Cloud automatically reduces costs." Wrong. Poorly designed workloads on auto-scaling platforms can become significantly more expensive. Monitor warehouse utilization, query patterns, storage growth, and auto-suspend policies from day one.

Architecture Principles I Follow

ARCHITECTURE PRINCIPLES FOR CLOUD MIGRATION PRINCIPLE 1 Separate Compute from Storage Don't replicate legacy patterns PRINCIPLE 2 ELT over ETL Push compute closer to data Transform in warehouse PRINCIPLE 3 Data as Product Owner · SLA Documentation Quality Metrics PRINCIPLE 4 Build for AI from Day One RAG · Agents · ML Semantic Search FUTURE CONSUMERS (Architect for All, Not Just Dashboards) Dashboards ML Models RAG Systems Agents Data Apps APIs Search If you only build for today's reports, you're already limiting tomorrow's capabilities
Fig 2: Four non-negotiable architecture principles + the full spectrum of future data consumers

Migration Do's & Don'ts

Do's

Don'ts

The Three Questions

Before migrating any object, ask:

  1. Does this data still provide business value?
  2. Would I design it this way if starting from scratch today?
  3. Will this support AI and analytics requirements over the next five years?

If the answer is "No" to any of these — redesign before migration.

Migration Readiness Scorecard

📋 Rate Your Organization (1-5 each)

0–15: Not Ready 16–25: Proceed with Caution 26–35: Migration Ready 36–40: Transformation Ready

Final Thoughts

The most successful migrations I've been part of were never database migrations. They were platform transformations.

The organizations that realize the highest value use migration as an opportunity to:

The cloud platform is excellent. But success is rarely determined by the technology. Success is determined by the architecture decisions made before the first table is migrated.

And as architects, that is where our focus should always remain.
Cloud MigrationSnowflakeData ArchitectureEnterpriseETL/ELTModernization