Introduction
Migrating Oracle Databases to the cloud isn’t just about moving data — it’s about re-engineering performance, reliability, and scalability.
Enterprises running mission-critical workloads like SAP HANA or Oracle Exadata need to ensure that every I/O, query, and transaction behaves just as efficiently — or better — in the new environment.
In this case study, we explore a real-world Oracle-to-Azure Cloud migration, focusing on storage performance validation, throughput optimization, and lessons learned in achieving consistent, high-performance workloads using Azure’s SSD and Ultra Disk tiers.
Why Azure Cloud for Oracle Databases?
Azure’s enterprise-grade infrastructure offers flexibility, strong SLAs, and integration with hybrid and multicloud architectures.
Organizations migrating Oracle workloads to Azure typically seek:
- Performance parity with on-prem Exadata or high-end storage.
- Elastic scalability for seasonal database workloads.
- Cost optimization through tiered storage and VM configurations.
- Integration with analytics and AI platforms natively available in Azure.
But the transition requires careful validation — particularly around I/O throughput, latency, and storage configuration.
The Real-World Scenario
A large enterprise running multiple Oracle databases on-premise decided to migrate to Azure Virtual Machines (DB VMs).
They initially provisioned Premium SSD disks, assuming the performance would match their on-prem infrastructure.
However, after deployment, they observed that the I/O response time during workload replay was significantly higher than during initial capture.
Upon review, the cause became clear — the Premium SSD tier had a maximum throughput of only 900 MB/s, insufficient for their I/O-intensive OLTP workloads.
Storage Options Evaluated
Here’s how different Azure storage tiers compared for Oracle database workloads:
| Disk Type | Scenario | Max Disk Size | Max Throughput | Max IOPS | Usable as OS Disk? |
|---|---|---|---|---|---|
| Ultra Disk | I/O-intensive workloads (SAP HANA, SQL, Oracle) | 65,536 GiB | 10,000 MB/s | 400,000 | ❌ |
| Premium SSD v2 | High-performance production workloads | 65,536 GiB | 1,200 MB/s | 80,000 | ❌ |
| Premium SSD | General production workloads | 32,767 GiB | 900 MB/s | 20,000 | ✅ |
| Standard SSD | Dev/Test, lightly used apps | 32,767 GiB | 750 MB/s | 6,000 | ✅ |
| Standard HDD | Backup, infrequent access | 32,767 GiB | 500 MB/s | 2,000 | ✅ |
Identifying the Performance Gap
The migration team noticed that I/O latency increased during replay tests using Database Replay (RAT) compared to capture in the on-prem environment.
The culprit was storage bottlenecking — the chosen Premium SSD simply couldn’t deliver the same throughput as required by Oracle’s redo and undo-intensive operations.
Solution: Provisioning Azure Ultra Disk
After analyzing workload characteristics (high read/write concurrency and frequent redo writes), the team upgraded the database storage to Azure Ultra Disk.
Benefits of Ultra Disk:
- Up to 10,000 MB/s throughput.
- Consistent low latency under 1 ms.
- Up to 400,000 IOPS.
- Adjustable performance levels without downtime.
Once the Ultra Disk was implemented, I/O response times matched or exceeded on-prem performance, and replay tests stabilized.
Performance Validation Strategy
The DBAs employed a structured performance validation process, using Oracle’s Real Application Testing (RAT) tools:
- Database Replay Capture:
Captured real production workload from the on-prem Oracle database. - Replay on Azure:
Replayed identical workloads in the Azure environment usingwrcclients. - Metric Comparison:
Evaluated throughput, IOPS, and SQL response times using AWR and OEM dashboards. - Storage-Level Validation:
Measured actual disk latency and throughput usingiostatandfioutilities within Azure VMs.
Result:
Post-migration replay metrics achieved 100% performance equivalence compared to on-prem values, proving Azure Ultra Disk suitable for enterprise-grade Oracle workloads.
Key Observations
- IOPS scaling: Performance linearly improved when scaling Ultra Disk sizes.
- Consistent throughput: Even under heavy concurrent DML operations.
- No jitter in response times — ideal for transaction-heavy systems.
- VM size matters: Azure VM types (e.g., Esv5-series) provided optimal disk bandwidth matching.
Security and Availability Enhancements
Alongside performance optimization, the migration included:
- Transparent Data Encryption (TDE) for Oracle database protection.
- Azure Disk Encryption using customer-managed keys (CMK).
- High Availability (HA) setup across Azure Availability Zones using Oracle Data Guard.
This ensured compliance, fault tolerance, and enterprise-grade security.
Lessons Learned
- Always benchmark before migration:
Don’t assume default disk tiers will meet I/O demands. - Use Ultra Disk for Oracle OLTP systems:
Especially when redo-intensive or high DML workloads exist. - Measure latency, not just IOPS:
Sub-millisecond consistency is vital for Oracle transaction engines. - Validate performance using Oracle RAT or AWR comparison reports.
- Adjust Azure VM sizing:
CPU, memory, and storage bandwidth must be tuned together.
Final Recommendation
For enterprises planning Oracle migrations to Azure Cloud:
- Use Premium SSD v2 for mid-range production workloads.
- Use Ultra Disk for high-performance transactional systems like Oracle, SAP HANA, or SQL Server.
- Employ Oracle Real Application Testing (RAT) to validate workloads before go-live.
- Implement a rollback plan (e.g., Oracle Data Guard or GoldenGate) for business continuity.
This approach ensures that Azure Cloud delivers enterprise-class performance without sacrificing Oracle’s database integrity or reliability.
Conclusion
Migrating Oracle databases to Azure Cloud isn’t a lift-and-shift — it’s a performance engineering exercise.
Through meticulous testing, proper storage selection, and validation using Oracle’s own tools, enterprises can achieve zero-performance-loss migrations while gaining the scalability and agility of the cloud.
Azure Ultra Disk proved to be the key differentiator — bridging the gap between cloud convenience and on-prem performance.
Explore further: Inside Oracle Database 23ai: Proven Upgrade, Migration & Performance Validation Strategies from Real-World Enterprises
Comments 2