The sorting logic, workload by workload, extends the migration posts’ dispositions. Oracle Database@Azure wins where the Oracle features are the point: RAC availability, Data Guard topologies, Exadata performance characteristics the workload measurably uses, and the vendor certified stacks (the ERP and banking cores) whose support matrices name Exadata. Migration to PostgreSQL (the September paths) wins where the database is Oracle by history rather than dependency, the license and lock in savings funding the conversion, and the fraction of estates in that category is larger than DBA folklore admits, which the assessment tooling quantifies per schema. Oracle on Azure VMs remains the niche middle for modest databases needing OS control without Exadata, carrying the licensing care it always did. And staying on premises stops being defensible once the same Exadata exists in region, the datacenter exit case from this week loses its last anchor tenant. Licensing threads through every branch: BYOL versus license included, the support rewards programs, and the commitment mathematics deserve a spreadsheet with finance in the room, because the numbers move seven figures on large estates.

Landing It Well

Treat the service as a first class tenant of the landing zone: delegated subnets planned in the hub and spoke address space with room for growth, the application spokes peered on the shortest path, DR designed with Data Guard across zones or to a second region’s deployment and rehearsed with the DR posts’ discipline, backups on Oracle’s tooling but restore drills on your calendar, and the seam instrumented, latency between app tier and database as a standing SLO, because the multicloud boundary is exactly where finger pointing lives during incidents and only your own measurements arbitrate. Migration onto it from on premises is Data Guard or RMAN based, the smoothest path in the whole migration catalog since both ends are Oracle. The strategic frame: this offering removes the infrastructure excuse from the Oracle conversation, leaving only the real question, which workloads are Oracle by requirement and which by inertia, and that question now gets answered per workload with a credible landing spot for both answers.

Cheers
Osama

The Decision Against the Alternatives

The sorting logic, workload by workload, extends the migration posts’ dispositions. Oracle Database@Azure wins where the Oracle features are the point: RAC availability, Data Guard topologies, Exadata performance characteristics the workload measurably uses, and the vendor certified stacks (the ERP and banking cores) whose support matrices name Exadata. Migration to PostgreSQL (the September paths) wins where the database is Oracle by history rather than dependency, the license and lock in savings funding the conversion, and the fraction of estates in that category is larger than DBA folklore admits, which the assessment tooling quantifies per schema. Oracle on Azure VMs remains the niche middle for modest databases needing OS control without Exadata, carrying the licensing care it always did. And staying on premises stops being defensible once the same Exadata exists in region, the datacenter exit case from this week loses its last anchor tenant. Licensing threads through every branch: BYOL versus license included, the support rewards programs, and the commitment mathematics deserve a spreadsheet with finance in the room, because the numbers move seven figures on large estates.

Landing It Well

Treat the service as a first class tenant of the landing zone: delegated subnets planned in the hub and spoke address space with room for growth, the application spokes peered on the shortest path, DR designed with Data Guard across zones or to a second region’s deployment and rehearsed with the DR posts’ discipline, backups on Oracle’s tooling but restore drills on your calendar, and the seam instrumented, latency between app tier and database as a standing SLO, because the multicloud boundary is exactly where finger pointing lives during incidents and only your own measurements arbitrate. Migration onto it from on premises is Data Guard or RMAN based, the smoothest path in the whole migration catalog since both ends are Oracle. The strategic frame: this offering removes the infrastructure excuse from the Oracle conversation, leaving only the real question, which workloads are Oracle by requirement and which by inertia, and that question now gets answered per workload with a credible landing spot for both answers.

Cheers
Osama

The offering places OCI managed infrastructure in Azure regions: Exadata Database Service on dedicated infrastructure for the full control RAC estate, and Autonomous Database for the workloads that fit its managed model, both provisioned through Azure portal and APIs, billed through your Azure agreement (counting toward Azure commitments, which the FinOps posts recognize as real money), and reachable from your VNets through delegated subnets with single digit millisecond latency to your application tiers. The operational split to internalize: Oracle runs the database infrastructure with OCI tooling and SLAs, your Azure estate provides the applications, networking, identity federation, and monitoring integration, and the multicloud seam, two operational models touching, is where the design attention goes. Entra integrates for identity, diagnostics and metrics flow toward your Log Analytics through the integration paths, and the network design keeps database client traffic inside the region’s fabric rather than hairpinning anywhere.

The Decision Against the Alternatives

The sorting logic, workload by workload, extends the migration posts’ dispositions. Oracle Database@Azure wins where the Oracle features are the point: RAC availability, Data Guard topologies, Exadata performance characteristics the workload measurably uses, and the vendor certified stacks (the ERP and banking cores) whose support matrices name Exadata. Migration to PostgreSQL (the September paths) wins where the database is Oracle by history rather than dependency, the license and lock in savings funding the conversion, and the fraction of estates in that category is larger than DBA folklore admits, which the assessment tooling quantifies per schema. Oracle on Azure VMs remains the niche middle for modest databases needing OS control without Exadata, carrying the licensing care it always did. And staying on premises stops being defensible once the same Exadata exists in region, the datacenter exit case from this week loses its last anchor tenant. Licensing threads through every branch: BYOL versus license included, the support rewards programs, and the commitment mathematics deserve a spreadsheet with finance in the room, because the numbers move seven figures on large estates.

Landing It Well

Treat the service as a first class tenant of the landing zone: delegated subnets planned in the hub and spoke address space with room for growth, the application spokes peered on the shortest path, DR designed with Data Guard across zones or to a second region’s deployment and rehearsed with the DR posts’ discipline, backups on Oracle’s tooling but restore drills on your calendar, and the seam instrumented, latency between app tier and database as a standing SLO, because the multicloud boundary is exactly where finger pointing lives during incidents and only your own measurements arbitrate. Migration onto it from on premises is Data Guard or RMAN based, the smoothest path in the whole migration catalog since both ends are Oracle. The strategic frame: this offering removes the infrastructure excuse from the Oracle conversation, leaving only the real question, which workloads are Oracle by requirement and which by inertia, and that question now gets answered per workload with a credible landing spot for both answers.

Cheers
Osama

The sorting logic, workload by workload, extends the migration posts’ dispositions. Oracle Database@Azure wins where the Oracle features are the point: RAC availability, Data Guard topologies, Exadata performance characteristics the workload measurably uses, and the vendor certified stacks (the ERP and banking cores) whose support matrices name Exadata. Migration to PostgreSQL (the September paths) wins where the database is Oracle by history rather than dependency, the license and lock in savings funding the conversion, and the fraction of estates in that category is larger than DBA folklore admits, which the assessment tooling quantifies per schema. Oracle on Azure VMs remains the niche middle for modest databases needing OS control without Exadata, carrying the licensing care it always did. And staying on premises stops being defensible once the same Exadata exists in region, the datacenter exit case from this week loses its last anchor tenant. Licensing threads through every branch: BYOL versus license included, the support rewards programs, and the commitment mathematics deserve a spreadsheet with finance in the room, because the numbers move seven figures on large estates.

Landing It Well

Treat the service as a first class tenant of the landing zone: delegated subnets planned in the hub and spoke address space with room for growth, the application spokes peered on the shortest path, DR designed with Data Guard across zones or to a second region’s deployment and rehearsed with the DR posts’ discipline, backups on Oracle’s tooling but restore drills on your calendar, and the seam instrumented, latency between app tier and database as a standing SLO, because the multicloud boundary is exactly where finger pointing lives during incidents and only your own measurements arbitrate. Migration onto it from on premises is Data Guard or RMAN based, the smoothest path in the whole migration catalog since both ends are Oracle. The strategic frame: this offering removes the infrastructure excuse from the Oracle conversation, leaving only the real question, which workloads are Oracle by requirement and which by inertia, and that question now gets answered per workload with a credible landing spot for both answers.

Cheers
Osama

The offering places OCI managed infrastructure in Azure regions: Exadata Database Service on dedicated infrastructure for the full control RAC estate, and Autonomous Database for the workloads that fit its managed model, both provisioned through Azure portal and APIs, billed through your Azure agreement (counting toward Azure commitments, which the FinOps posts recognize as real money), and reachable from your VNets through delegated subnets with single digit millisecond latency to your application tiers. The operational split to internalize: Oracle runs the database infrastructure with OCI tooling and SLAs, your Azure estate provides the applications, networking, identity federation, and monitoring integration, and the multicloud seam, two operational models touching, is where the design attention goes. Entra integrates for identity, diagnostics and metrics flow toward your Log Analytics through the integration paths, and the network design keeps database client traffic inside the region’s fabric rather than hairpinning anywhere.

The Decision Against the Alternatives

The sorting logic, workload by workload, extends the migration posts’ dispositions. Oracle Database@Azure wins where the Oracle features are the point: RAC availability, Data Guard topologies, Exadata performance characteristics the workload measurably uses, and the vendor certified stacks (the ERP and banking cores) whose support matrices name Exadata. Migration to PostgreSQL (the September paths) wins where the database is Oracle by history rather than dependency, the license and lock in savings funding the conversion, and the fraction of estates in that category is larger than DBA folklore admits, which the assessment tooling quantifies per schema. Oracle on Azure VMs remains the niche middle for modest databases needing OS control without Exadata, carrying the licensing care it always did. And staying on premises stops being defensible once the same Exadata exists in region, the datacenter exit case from this week loses its last anchor tenant. Licensing threads through every branch: BYOL versus license included, the support rewards programs, and the commitment mathematics deserve a spreadsheet with finance in the room, because the numbers move seven figures on large estates.

Landing It Well

Treat the service as a first class tenant of the landing zone: delegated subnets planned in the hub and spoke address space with room for growth, the application spokes peered on the shortest path, DR designed with Data Guard across zones or to a second region’s deployment and rehearsed with the DR posts’ discipline, backups on Oracle’s tooling but restore drills on your calendar, and the seam instrumented, latency between app tier and database as a standing SLO, because the multicloud boundary is exactly where finger pointing lives during incidents and only your own measurements arbitrate. Migration onto it from on premises is Data Guard or RMAN based, the smoothest path in the whole migration catalog since both ends are Oracle. The strategic frame: this offering removes the infrastructure excuse from the Oracle conversation, leaving only the real question, which workloads are Oracle by requirement and which by inertia, and that question now gets answered per workload with a credible landing spot for both answers.

Cheers
Osama

The Decision Against the Alternatives

The sorting logic, workload by workload, extends the migration posts’ dispositions. Oracle Database@Azure wins where the Oracle features are the point: RAC availability, Data Guard topologies, Exadata performance characteristics the workload measurably uses, and the vendor certified stacks (the ERP and banking cores) whose support matrices name Exadata. Migration to PostgreSQL (the September paths) wins where the database is Oracle by history rather than dependency, the license and lock in savings funding the conversion, and the fraction of estates in that category is larger than DBA folklore admits, which the assessment tooling quantifies per schema. Oracle on Azure VMs remains the niche middle for modest databases needing OS control without Exadata, carrying the licensing care it always did. And staying on premises stops being defensible once the same Exadata exists in region, the datacenter exit case from this week loses its last anchor tenant. Licensing threads through every branch: BYOL versus license included, the support rewards programs, and the commitment mathematics deserve a spreadsheet with finance in the room, because the numbers move seven figures on large estates.

Landing It Well

Treat the service as a first class tenant of the landing zone: delegated subnets planned in the hub and spoke address space with room for growth, the application spokes peered on the shortest path, DR designed with Data Guard across zones or to a second region’s deployment and rehearsed with the DR posts’ discipline, backups on Oracle’s tooling but restore drills on your calendar, and the seam instrumented, latency between app tier and database as a standing SLO, because the multicloud boundary is exactly where finger pointing lives during incidents and only your own measurements arbitrate. Migration onto it from on premises is Data Guard or RMAN based, the smoothest path in the whole migration catalog since both ends are Oracle. The strategic frame: this offering removes the infrastructure excuse from the Oracle conversation, leaving only the real question, which workloads are Oracle by requirement and which by inertia, and that question now gets answered per workload with a credible landing spot for both answers.

Cheers
Osama

The offering places OCI managed infrastructure in Azure regions: Exadata Database Service on dedicated infrastructure for the full control RAC estate, and Autonomous Database for the workloads that fit its managed model, both provisioned through Azure portal and APIs, billed through your Azure agreement (counting toward Azure commitments, which the FinOps posts recognize as real money), and reachable from your VNets through delegated subnets with single digit millisecond latency to your application tiers. The operational split to internalize: Oracle runs the database infrastructure with OCI tooling and SLAs, your Azure estate provides the applications, networking, identity federation, and monitoring integration, and the multicloud seam, two operational models touching, is where the design attention goes. Entra integrates for identity, diagnostics and metrics flow toward your Log Analytics through the integration paths, and the network design keeps database client traffic inside the region’s fabric rather than hairpinning anywhere.

The Decision Against the Alternatives

The sorting logic, workload by workload, extends the migration posts’ dispositions. Oracle Database@Azure wins where the Oracle features are the point: RAC availability, Data Guard topologies, Exadata performance characteristics the workload measurably uses, and the vendor certified stacks (the ERP and banking cores) whose support matrices name Exadata. Migration to PostgreSQL (the September paths) wins where the database is Oracle by history rather than dependency, the license and lock in savings funding the conversion, and the fraction of estates in that category is larger than DBA folklore admits, which the assessment tooling quantifies per schema. Oracle on Azure VMs remains the niche middle for modest databases needing OS control without Exadata, carrying the licensing care it always did. And staying on premises stops being defensible once the same Exadata exists in region, the datacenter exit case from this week loses its last anchor tenant. Licensing threads through every branch: BYOL versus license included, the support rewards programs, and the commitment mathematics deserve a spreadsheet with finance in the room, because the numbers move seven figures on large estates.

Landing It Well

Treat the service as a first class tenant of the landing zone: delegated subnets planned in the hub and spoke address space with room for growth, the application spokes peered on the shortest path, DR designed with Data Guard across zones or to a second region’s deployment and rehearsed with the DR posts’ discipline, backups on Oracle’s tooling but restore drills on your calendar, and the seam instrumented, latency between app tier and database as a standing SLO, because the multicloud boundary is exactly where finger pointing lives during incidents and only your own measurements arbitrate. Migration onto it from on premises is Data Guard or RMAN based, the smoothest path in the whole migration catalog since both ends are Oracle. The strategic frame: this offering removes the infrastructure excuse from the Oracle conversation, leaving only the real question, which workloads are Oracle by requirement and which by inertia, and that question now gets answered per workload with a credible landing spot for both answers.

Cheers
Osama

The Decision Against the Alternatives

The sorting logic, workload by workload, extends the migration posts’ dispositions. Oracle Database@Azure wins where the Oracle features are the point: RAC availability, Data Guard topologies, Exadata performance characteristics the workload measurably uses, and the vendor certified stacks (the ERP and banking cores) whose support matrices name Exadata. Migration to PostgreSQL (the September paths) wins where the database is Oracle by history rather than dependency, the license and lock in savings funding the conversion, and the fraction of estates in that category is larger than DBA folklore admits, which the assessment tooling quantifies per schema. Oracle on Azure VMs remains the niche middle for modest databases needing OS control without Exadata, carrying the licensing care it always did. And staying on premises stops being defensible once the same Exadata exists in region, the datacenter exit case from this week loses its last anchor tenant. Licensing threads through every branch: BYOL versus license included, the support rewards programs, and the commitment mathematics deserve a spreadsheet with finance in the room, because the numbers move seven figures on large estates.

Landing It Well

Treat the service as a first class tenant of the landing zone: delegated subnets planned in the hub and spoke address space with room for growth, the application spokes peered on the shortest path, DR designed with Data Guard across zones or to a second region’s deployment and rehearsed with the DR posts’ discipline, backups on Oracle’s tooling but restore drills on your calendar, and the seam instrumented, latency between app tier and database as a standing SLO, because the multicloud boundary is exactly where finger pointing lives during incidents and only your own measurements arbitrate. Migration onto it from on premises is Data Guard or RMAN based, the smoothest path in the whole migration catalog since both ends are Oracle. The strategic frame: this offering removes the infrastructure excuse from the Oracle conversation, leaving only the real question, which workloads are Oracle by requirement and which by inertia, and that question now gets answered per workload with a credible landing spot for both answers.

Cheers
Osama

The offering places OCI managed infrastructure in Azure regions: Exadata Database Service on dedicated infrastructure for the full control RAC estate, and Autonomous Database for the workloads that fit its managed model, both provisioned through Azure portal and APIs, billed through your Azure agreement (counting toward Azure commitments, which the FinOps posts recognize as real money), and reachable from your VNets through delegated subnets with single digit millisecond latency to your application tiers. The operational split to internalize: Oracle runs the database infrastructure with OCI tooling and SLAs, your Azure estate provides the applications, networking, identity federation, and monitoring integration, and the multicloud seam, two operational models touching, is where the design attention goes. Entra integrates for identity, diagnostics and metrics flow toward your Log Analytics through the integration paths, and the network design keeps database client traffic inside the region’s fabric rather than hairpinning anywhere.

The Decision Against the Alternatives

The sorting logic, workload by workload, extends the migration posts’ dispositions. Oracle Database@Azure wins where the Oracle features are the point: RAC availability, Data Guard topologies, Exadata performance characteristics the workload measurably uses, and the vendor certified stacks (the ERP and banking cores) whose support matrices name Exadata. Migration to PostgreSQL (the September paths) wins where the database is Oracle by history rather than dependency, the license and lock in savings funding the conversion, and the fraction of estates in that category is larger than DBA folklore admits, which the assessment tooling quantifies per schema. Oracle on Azure VMs remains the niche middle for modest databases needing OS control without Exadata, carrying the licensing care it always did. And staying on premises stops being defensible once the same Exadata exists in region, the datacenter exit case from this week loses its last anchor tenant. Licensing threads through every branch: BYOL versus license included, the support rewards programs, and the commitment mathematics deserve a spreadsheet with finance in the room, because the numbers move seven figures on large estates.

Landing It Well

Treat the service as a first class tenant of the landing zone: delegated subnets planned in the hub and spoke address space with room for growth, the application spokes peered on the shortest path, DR designed with Data Guard across zones or to a second region’s deployment and rehearsed with the DR posts’ discipline, backups on Oracle’s tooling but restore drills on your calendar, and the seam instrumented, latency between app tier and database as a standing SLO, because the multicloud boundary is exactly where finger pointing lives during incidents and only your own measurements arbitrate. Migration onto it from on premises is Data Guard or RMAN based, the smoothest path in the whole migration catalog since both ends are Oracle. The strategic frame: this offering removes the infrastructure excuse from the Oracle conversation, leaving only the real question, which workloads are Oracle by requirement and which by inertia, and that question now gets answered per workload with a credible landing spot for both answers.

Cheers
Osama

The sorting logic, workload by workload, extends the migration posts’ dispositions. Oracle Database@Azure wins where the Oracle features are the point: RAC availability, Data Guard topologies, Exadata performance characteristics the workload measurably uses, and the vendor certified stacks (the ERP and banking cores) whose support matrices name Exadata. Migration to PostgreSQL (the September paths) wins where the database is Oracle by history rather than dependency, the license and lock in savings funding the conversion, and the fraction of estates in that category is larger than DBA folklore admits, which the assessment tooling quantifies per schema. Oracle on Azure VMs remains the niche middle for modest databases needing OS control without Exadata, carrying the licensing care it always did. And staying on premises stops being defensible once the same Exadata exists in region, the datacenter exit case from this week loses its last anchor tenant. Licensing threads through every branch: BYOL versus license included, the support rewards programs, and the commitment mathematics deserve a spreadsheet with finance in the room, because the numbers move seven figures on large estates.

Landing It Well

Treat the service as a first class tenant of the landing zone: delegated subnets planned in the hub and spoke address space with room for growth, the application spokes peered on the shortest path, DR designed with Data Guard across zones or to a second region’s deployment and rehearsed with the DR posts’ discipline, backups on Oracle’s tooling but restore drills on your calendar, and the seam instrumented, latency between app tier and database as a standing SLO, because the multicloud boundary is exactly where finger pointing lives during incidents and only your own measurements arbitrate. Migration onto it from on premises is Data Guard or RMAN based, the smoothest path in the whole migration catalog since both ends are Oracle. The strategic frame: this offering removes the infrastructure excuse from the Oracle conversation, leaving only the real question, which workloads are Oracle by requirement and which by inertia, and that question now gets answered per workload with a credible landing spot for both answers.

Cheers
Osama

The Decision Against the Alternatives

The sorting logic, workload by workload, extends the migration posts’ dispositions. Oracle Database@Azure wins where the Oracle features are the point: RAC availability, Data Guard topologies, Exadata performance characteristics the workload measurably uses, and the vendor certified stacks (the ERP and banking cores) whose support matrices name Exadata. Migration to PostgreSQL (the September paths) wins where the database is Oracle by history rather than dependency, the license and lock in savings funding the conversion, and the fraction of estates in that category is larger than DBA folklore admits, which the assessment tooling quantifies per schema. Oracle on Azure VMs remains the niche middle for modest databases needing OS control without Exadata, carrying the licensing care it always did. And staying on premises stops being defensible once the same Exadata exists in region, the datacenter exit case from this week loses its last anchor tenant. Licensing threads through every branch: BYOL versus license included, the support rewards programs, and the commitment mathematics deserve a spreadsheet with finance in the room, because the numbers move seven figures on large estates.

Landing It Well

Treat the service as a first class tenant of the landing zone: delegated subnets planned in the hub and spoke address space with room for growth, the application spokes peered on the shortest path, DR designed with Data Guard across zones or to a second region’s deployment and rehearsed with the DR posts’ discipline, backups on Oracle’s tooling but restore drills on your calendar, and the seam instrumented, latency between app tier and database as a standing SLO, because the multicloud boundary is exactly where finger pointing lives during incidents and only your own measurements arbitrate. Migration onto it from on premises is Data Guard or RMAN based, the smoothest path in the whole migration catalog since both ends are Oracle. The strategic frame: this offering removes the infrastructure excuse from the Oracle conversation, leaving only the real question, which workloads are Oracle by requirement and which by inertia, and that question now gets answered per workload with a credible landing spot for both answers.

Cheers
Osama

The offering places OCI managed infrastructure in Azure regions: Exadata Database Service on dedicated infrastructure for the full control RAC estate, and Autonomous Database for the workloads that fit its managed model, both provisioned through Azure portal and APIs, billed through your Azure agreement (counting toward Azure commitments, which the FinOps posts recognize as real money), and reachable from your VNets through delegated subnets with single digit millisecond latency to your application tiers. The operational split to internalize: Oracle runs the database infrastructure with OCI tooling and SLAs, your Azure estate provides the applications, networking, identity federation, and monitoring integration, and the multicloud seam, two operational models touching, is where the design attention goes. Entra integrates for identity, diagnostics and metrics flow toward your Log Analytics through the integration paths, and the network design keeps database client traffic inside the region’s fabric rather than hairpinning anywhere.

The Decision Against the Alternatives

The sorting logic, workload by workload, extends the migration posts’ dispositions. Oracle Database@Azure wins where the Oracle features are the point: RAC availability, Data Guard topologies, Exadata performance characteristics the workload measurably uses, and the vendor certified stacks (the ERP and banking cores) whose support matrices name Exadata. Migration to PostgreSQL (the September paths) wins where the database is Oracle by history rather than dependency, the license and lock in savings funding the conversion, and the fraction of estates in that category is larger than DBA folklore admits, which the assessment tooling quantifies per schema. Oracle on Azure VMs remains the niche middle for modest databases needing OS control without Exadata, carrying the licensing care it always did. And staying on premises stops being defensible once the same Exadata exists in region, the datacenter exit case from this week loses its last anchor tenant. Licensing threads through every branch: BYOL versus license included, the support rewards programs, and the commitment mathematics deserve a spreadsheet with finance in the room, because the numbers move seven figures on large estates.

Landing It Well

Treat the service as a first class tenant of the landing zone: delegated subnets planned in the hub and spoke address space with room for growth, the application spokes peered on the shortest path, DR designed with Data Guard across zones or to a second region’s deployment and rehearsed with the DR posts’ discipline, backups on Oracle’s tooling but restore drills on your calendar, and the seam instrumented, latency between app tier and database as a standing SLO, because the multicloud boundary is exactly where finger pointing lives during incidents and only your own measurements arbitrate. Migration onto it from on premises is Data Guard or RMAN based, the smoothest path in the whole migration catalog since both ends are Oracle. The strategic frame: this offering removes the infrastructure excuse from the Oracle conversation, leaving only the real question, which workloads are Oracle by requirement and which by inertia, and that question now gets answered per workload with a credible landing spot for both answers.

Cheers
Osama

The Decision Against the Alternatives

The sorting logic, workload by workload, extends the migration posts’ dispositions. Oracle Database@Azure wins where the Oracle features are the point: RAC availability, Data Guard topologies, Exadata performance characteristics the workload measurably uses, and the vendor certified stacks (the ERP and banking cores) whose support matrices name Exadata. Migration to PostgreSQL (the September paths) wins where the database is Oracle by history rather than dependency, the license and lock in savings funding the conversion, and the fraction of estates in that category is larger than DBA folklore admits, which the assessment tooling quantifies per schema. Oracle on Azure VMs remains the niche middle for modest databases needing OS control without Exadata, carrying the licensing care it always did. And staying on premises stops being defensible once the same Exadata exists in region, the datacenter exit case from this week loses its last anchor tenant. Licensing threads through every branch: BYOL versus license included, the support rewards programs, and the commitment mathematics deserve a spreadsheet with finance in the room, because the numbers move seven figures on large estates.

Landing It Well

Treat the service as a first class tenant of the landing zone: delegated subnets planned in the hub and spoke address space with room for growth, the application spokes peered on the shortest path, DR designed with Data Guard across zones or to a second region’s deployment and rehearsed with the DR posts’ discipline, backups on Oracle’s tooling but restore drills on your calendar, and the seam instrumented, latency between app tier and database as a standing SLO, because the multicloud boundary is exactly where finger pointing lives during incidents and only your own measurements arbitrate. Migration onto it from on premises is Data Guard or RMAN based, the smoothest path in the whole migration catalog since both ends are Oracle. The strategic frame: this offering removes the infrastructure excuse from the Oracle conversation, leaving only the real question, which workloads are Oracle by requirement and which by inertia, and that question now gets answered per workload with a credible landing spot for both answers.

Cheers
Osama

The Decision Against the Alternatives

The sorting logic, workload by workload, extends the migration posts’ dispositions. Oracle Database@Azure wins where the Oracle features are the point: RAC availability, Data Guard topologies, Exadata performance characteristics the workload measurably uses, and the vendor certified stacks (the ERP and banking cores) whose support matrices name Exadata. Migration to PostgreSQL (the September paths) wins where the database is Oracle by history rather than dependency, the license and lock in savings funding the conversion, and the fraction of estates in that category is larger than DBA folklore admits, which the assessment tooling quantifies per schema. Oracle on Azure VMs remains the niche middle for modest databases needing OS control without Exadata, carrying the licensing care it always did. And staying on premises stops being defensible once the same Exadata exists in region, the datacenter exit case from this week loses its last anchor tenant. Licensing threads through every branch: BYOL versus license included, the support rewards programs, and the commitment mathematics deserve a spreadsheet with finance in the room, because the numbers move seven figures on large estates.

Landing It Well

Treat the service as a first class tenant of the landing zone: delegated subnets planned in the hub and spoke address space with room for growth, the application spokes peered on the shortest path, DR designed with Data Guard across zones or to a second region’s deployment and rehearsed with the DR posts’ discipline, backups on Oracle’s tooling but restore drills on your calendar, and the seam instrumented, latency between app tier and database as a standing SLO, because the multicloud boundary is exactly where finger pointing lives during incidents and only your own measurements arbitrate. Migration onto it from on premises is Data Guard or RMAN based, the smoothest path in the whole migration catalog since both ends are Oracle. The strategic frame: this offering removes the infrastructure excuse from the Oracle conversation, leaving only the real question, which workloads are Oracle by requirement and which by inertia, and that question now gets answered per workload with a credible landing spot for both answers.

Cheers
Osama

The offering places OCI managed infrastructure in Azure regions: Exadata Database Service on dedicated infrastructure for the full control RAC estate, and Autonomous Database for the workloads that fit its managed model, both provisioned through Azure portal and APIs, billed through your Azure agreement (counting toward Azure commitments, which the FinOps posts recognize as real money), and reachable from your VNets through delegated subnets with single digit millisecond latency to your application tiers. The operational split to internalize: Oracle runs the database infrastructure with OCI tooling and SLAs, your Azure estate provides the applications, networking, identity federation, and monitoring integration, and the multicloud seam, two operational models touching, is where the design attention goes. Entra integrates for identity, diagnostics and metrics flow toward your Log Analytics through the integration paths, and the network design keeps database client traffic inside the region’s fabric rather than hairpinning anywhere.

The Decision Against the Alternatives

The sorting logic, workload by workload, extends the migration posts’ dispositions. Oracle Database@Azure wins where the Oracle features are the point: RAC availability, Data Guard topologies, Exadata performance characteristics the workload measurably uses, and the vendor certified stacks (the ERP and banking cores) whose support matrices name Exadata. Migration to PostgreSQL (the September paths) wins where the database is Oracle by history rather than dependency, the license and lock in savings funding the conversion, and the fraction of estates in that category is larger than DBA folklore admits, which the assessment tooling quantifies per schema. Oracle on Azure VMs remains the niche middle for modest databases needing OS control without Exadata, carrying the licensing care it always did. And staying on premises stops being defensible once the same Exadata exists in region, the datacenter exit case from this week loses its last anchor tenant. Licensing threads through every branch: BYOL versus license included, the support rewards programs, and the commitment mathematics deserve a spreadsheet with finance in the room, because the numbers move seven figures on large estates.

Landing It Well

Treat the service as a first class tenant of the landing zone: delegated subnets planned in the hub and spoke address space with room for growth, the application spokes peered on the shortest path, DR designed with Data Guard across zones or to a second region’s deployment and rehearsed with the DR posts’ discipline, backups on Oracle’s tooling but restore drills on your calendar, and the seam instrumented, latency between app tier and database as a standing SLO, because the multicloud boundary is exactly where finger pointing lives during incidents and only your own measurements arbitrate. Migration onto it from on premises is Data Guard or RMAN based, the smoothest path in the whole migration catalog since both ends are Oracle. The strategic frame: this offering removes the infrastructure excuse from the Oracle conversation, leaving only the real question, which workloads are Oracle by requirement and which by inertia, and that question now gets answered per workload with a credible landing spot for both answers.

Cheers
Osama

The sorting logic, workload by workload, extends the migration posts’ dispositions. Oracle Database@Azure wins where the Oracle features are the point: RAC availability, Data Guard topologies, Exadata performance characteristics the workload measurably uses, and the vendor certified stacks (the ERP and banking cores) whose support matrices name Exadata. Migration to PostgreSQL (the September paths) wins where the database is Oracle by history rather than dependency, the license and lock in savings funding the conversion, and the fraction of estates in that category is larger than DBA folklore admits, which the assessment tooling quantifies per schema. Oracle on Azure VMs remains the niche middle for modest databases needing OS control without Exadata, carrying the licensing care it always did. And staying on premises stops being defensible once the same Exadata exists in region, the datacenter exit case from this week loses its last anchor tenant. Licensing threads through every branch: BYOL versus license included, the support rewards programs, and the commitment mathematics deserve a spreadsheet with finance in the room, because the numbers move seven figures on large estates.

Landing It Well

Treat the service as a first class tenant of the landing zone: delegated subnets planned in the hub and spoke address space with room for growth, the application spokes peered on the shortest path, DR designed with Data Guard across zones or to a second region’s deployment and rehearsed with the DR posts’ discipline, backups on Oracle’s tooling but restore drills on your calendar, and the seam instrumented, latency between app tier and database as a standing SLO, because the multicloud boundary is exactly where finger pointing lives during incidents and only your own measurements arbitrate. Migration onto it from on premises is Data Guard or RMAN based, the smoothest path in the whole migration catalog since both ends are Oracle. The strategic frame: this offering removes the infrastructure excuse from the Oracle conversation, leaving only the real question, which workloads are Oracle by requirement and which by inertia, and that question now gets answered per workload with a credible landing spot for both answers.

Cheers
Osama

The Decision Against the Alternatives

The sorting logic, workload by workload, extends the migration posts’ dispositions. Oracle Database@Azure wins where the Oracle features are the point: RAC availability, Data Guard topologies, Exadata performance characteristics the workload measurably uses, and the vendor certified stacks (the ERP and banking cores) whose support matrices name Exadata. Migration to PostgreSQL (the September paths) wins where the database is Oracle by history rather than dependency, the license and lock in savings funding the conversion, and the fraction of estates in that category is larger than DBA folklore admits, which the assessment tooling quantifies per schema. Oracle on Azure VMs remains the niche middle for modest databases needing OS control without Exadata, carrying the licensing care it always did. And staying on premises stops being defensible once the same Exadata exists in region, the datacenter exit case from this week loses its last anchor tenant. Licensing threads through every branch: BYOL versus license included, the support rewards programs, and the commitment mathematics deserve a spreadsheet with finance in the room, because the numbers move seven figures on large estates.

Landing It Well

Treat the service as a first class tenant of the landing zone: delegated subnets planned in the hub and spoke address space with room for growth, the application spokes peered on the shortest path, DR designed with Data Guard across zones or to a second region’s deployment and rehearsed with the DR posts’ discipline, backups on Oracle’s tooling but restore drills on your calendar, and the seam instrumented, latency between app tier and database as a standing SLO, because the multicloud boundary is exactly where finger pointing lives during incidents and only your own measurements arbitrate. Migration onto it from on premises is Data Guard or RMAN based, the smoothest path in the whole migration catalog since both ends are Oracle. The strategic frame: this offering removes the infrastructure excuse from the Oracle conversation, leaving only the real question, which workloads are Oracle by requirement and which by inertia, and that question now gets answered per workload with a credible landing spot for both answers.

Cheers
Osama

The Decision Against the Alternatives

The sorting logic, workload by workload, extends the migration posts’ dispositions. Oracle Database@Azure wins where the Oracle features are the point: RAC availability, Data Guard topologies, Exadata performance characteristics the workload measurably uses, and the vendor certified stacks (the ERP and banking cores) whose support matrices name Exadata. Migration to PostgreSQL (the September paths) wins where the database is Oracle by history rather than dependency, the license and lock in savings funding the conversion, and the fraction of estates in that category is larger than DBA folklore admits, which the assessment tooling quantifies per schema. Oracle on Azure VMs remains the niche middle for modest databases needing OS control without Exadata, carrying the licensing care it always did. And staying on premises stops being defensible once the same Exadata exists in region, the datacenter exit case from this week loses its last anchor tenant. Licensing threads through every branch: BYOL versus license included, the support rewards programs, and the commitment mathematics deserve a spreadsheet with finance in the room, because the numbers move seven figures on large estates.

Landing It Well

Treat the service as a first class tenant of the landing zone: delegated subnets planned in the hub and spoke address space with room for growth, the application spokes peered on the shortest path, DR designed with Data Guard across zones or to a second region’s deployment and rehearsed with the DR posts’ discipline, backups on Oracle’s tooling but restore drills on your calendar, and the seam instrumented, latency between app tier and database as a standing SLO, because the multicloud boundary is exactly where finger pointing lives during incidents and only your own measurements arbitrate. Migration onto it from on premises is Data Guard or RMAN based, the smoothest path in the whole migration catalog since both ends are Oracle. The strategic frame: this offering removes the infrastructure excuse from the Oracle conversation, leaving only the real question, which workloads are Oracle by requirement and which by inertia, and that question now gets answered per workload with a credible landing spot for both answers.

Cheers
Osama

The offering places OCI managed infrastructure in Azure regions: Exadata Database Service on dedicated infrastructure for the full control RAC estate, and Autonomous Database for the workloads that fit its managed model, both provisioned through Azure portal and APIs, billed through your Azure agreement (counting toward Azure commitments, which the FinOps posts recognize as real money), and reachable from your VNets through delegated subnets with single digit millisecond latency to your application tiers. The operational split to internalize: Oracle runs the database infrastructure with OCI tooling and SLAs, your Azure estate provides the applications, networking, identity federation, and monitoring integration, and the multicloud seam, two operational models touching, is where the design attention goes. Entra integrates for identity, diagnostics and metrics flow toward your Log Analytics through the integration paths, and the network design keeps database client traffic inside the region’s fabric rather than hairpinning anywhere.

The Decision Against the Alternatives

The sorting logic, workload by workload, extends the migration posts’ dispositions. Oracle Database@Azure wins where the Oracle features are the point: RAC availability, Data Guard topologies, Exadata performance characteristics the workload measurably uses, and the vendor certified stacks (the ERP and banking cores) whose support matrices name Exadata. Migration to PostgreSQL (the September paths) wins where the database is Oracle by history rather than dependency, the license and lock in savings funding the conversion, and the fraction of estates in that category is larger than DBA folklore admits, which the assessment tooling quantifies per schema. Oracle on Azure VMs remains the niche middle for modest databases needing OS control without Exadata, carrying the licensing care it always did. And staying on premises stops being defensible once the same Exadata exists in region, the datacenter exit case from this week loses its last anchor tenant. Licensing threads through every branch: BYOL versus license included, the support rewards programs, and the commitment mathematics deserve a spreadsheet with finance in the room, because the numbers move seven figures on large estates.

Landing It Well

Treat the service as a first class tenant of the landing zone: delegated subnets planned in the hub and spoke address space with room for growth, the application spokes peered on the shortest path, DR designed with Data Guard across zones or to a second region’s deployment and rehearsed with the DR posts’ discipline, backups on Oracle’s tooling but restore drills on your calendar, and the seam instrumented, latency between app tier and database as a standing SLO, because the multicloud boundary is exactly where finger pointing lives during incidents and only your own measurements arbitrate. Migration onto it from on premises is Data Guard or RMAN based, the smoothest path in the whole migration catalog since both ends are Oracle. The strategic frame: this offering removes the infrastructure excuse from the Oracle conversation, leaving only the real question, which workloads are Oracle by requirement and which by inertia, and that question now gets answered per workload with a credible landing spot for both answers.

Cheers
Osama

Every large estate has the Oracle question, and for years the answers were all compromises: run it on Azure VMs without RAC and with licensing arithmetic that punished vCPUs, migrate off at PL/SQL conversion cost (the September migration post’s honest quarter of work), or keep the datacenter alive for one database. Oracle Database@Azure changed the option list: Oracle’s own Exadata and Autonomous services, operated by Oracle, physically deployed inside Azure datacenters, purchasable through Azure.

What It Actually Is

Treat the service as a first class tenant of the landing zone: delegated subnets planned in the hub and spoke address space with room for growth, the application spokes peered on the shortest path, DR designed with Data Guard across zones or to a second region’s deployment and rehearsed with the DR posts’ discipline, backups on Oracle’s tooling but restore drills on your calendar, and the seam instrumented, latency between app tier and database as a standing SLO, because the multicloud boundary is exactly where finger pointing lives during incidents and only your own measurements arbitrate. Migration onto it from on premises is Data Guard or RMAN based, the smoothest path in the whole migration catalog since both ends are Oracle. The strategic frame: this offering removes the infrastructure excuse from the Oracle conversation, leaving only the real question, which workloads are Oracle by requirement and which by inertia, and that question now gets answered per workload with a credible landing spot for both answers.

Cheers
Osama

The sorting logic, workload by workload, extends the migration posts’ dispositions. Oracle Database@Azure wins where the Oracle features are the point: RAC availability, Data Guard topologies, Exadata performance characteristics the workload measurably uses, and the vendor certified stacks (the ERP and banking cores) whose support matrices name Exadata. Migration to PostgreSQL (the September paths) wins where the database is Oracle by history rather than dependency, the license and lock in savings funding the conversion, and the fraction of estates in that category is larger than DBA folklore admits, which the assessment tooling quantifies per schema. Oracle on Azure VMs remains the niche middle for modest databases needing OS control without Exadata, carrying the licensing care it always did. And staying on premises stops being defensible once the same Exadata exists in region, the datacenter exit case from this week loses its last anchor tenant. Licensing threads through every branch: BYOL versus license included, the support rewards programs, and the commitment mathematics deserve a spreadsheet with finance in the room, because the numbers move seven figures on large estates.

Landing It Well

Treat the service as a first class tenant of the landing zone: delegated subnets planned in the hub and spoke address space with room for growth, the application spokes peered on the shortest path, DR designed with Data Guard across zones or to a second region’s deployment and rehearsed with the DR posts’ discipline, backups on Oracle’s tooling but restore drills on your calendar, and the seam instrumented, latency between app tier and database as a standing SLO, because the multicloud boundary is exactly where finger pointing lives during incidents and only your own measurements arbitrate. Migration onto it from on premises is Data Guard or RMAN based, the smoothest path in the whole migration catalog since both ends are Oracle. The strategic frame: this offering removes the infrastructure excuse from the Oracle conversation, leaving only the real question, which workloads are Oracle by requirement and which by inertia, and that question now gets answered per workload with a credible landing spot for both answers.

Cheers
Osama

The Decision Against the Alternatives

The sorting logic, workload by workload, extends the migration posts’ dispositions. Oracle Database@Azure wins where the Oracle features are the point: RAC availability, Data Guard topologies, Exadata performance characteristics the workload measurably uses, and the vendor certified stacks (the ERP and banking cores) whose support matrices name Exadata. Migration to PostgreSQL (the September paths) wins where the database is Oracle by history rather than dependency, the license and lock in savings funding the conversion, and the fraction of estates in that category is larger than DBA folklore admits, which the assessment tooling quantifies per schema. Oracle on Azure VMs remains the niche middle for modest databases needing OS control without Exadata, carrying the licensing care it always did. And staying on premises stops being defensible once the same Exadata exists in region, the datacenter exit case from this week loses its last anchor tenant. Licensing threads through every branch: BYOL versus license included, the support rewards programs, and the commitment mathematics deserve a spreadsheet with finance in the room, because the numbers move seven figures on large estates.

Landing It Well

Treat the service as a first class tenant of the landing zone: delegated subnets planned in the hub and spoke address space with room for growth, the application spokes peered on the shortest path, DR designed with Data Guard across zones or to a second region’s deployment and rehearsed with the DR posts’ discipline, backups on Oracle’s tooling but restore drills on your calendar, and the seam instrumented, latency between app tier and database as a standing SLO, because the multicloud boundary is exactly where finger pointing lives during incidents and only your own measurements arbitrate. Migration onto it from on premises is Data Guard or RMAN based, the smoothest path in the whole migration catalog since both ends are Oracle. The strategic frame: this offering removes the infrastructure excuse from the Oracle conversation, leaving only the real question, which workloads are Oracle by requirement and which by inertia, and that question now gets answered per workload with a credible landing spot for both answers.

Cheers
Osama

The Decision Against the Alternatives

The sorting logic, workload by workload, extends the migration posts’ dispositions. Oracle Database@Azure wins where the Oracle features are the point: RAC availability, Data Guard topologies, Exadata performance characteristics the workload measurably uses, and the vendor certified stacks (the ERP and banking cores) whose support matrices name Exadata. Migration to PostgreSQL (the September paths) wins where the database is Oracle by history rather than dependency, the license and lock in savings funding the conversion, and the fraction of estates in that category is larger than DBA folklore admits, which the assessment tooling quantifies per schema. Oracle on Azure VMs remains the niche middle for modest databases needing OS control without Exadata, carrying the licensing care it always did. And staying on premises stops being defensible once the same Exadata exists in region, the datacenter exit case from this week loses its last anchor tenant. Licensing threads through every branch: BYOL versus license included, the support rewards programs, and the commitment mathematics deserve a spreadsheet with finance in the room, because the numbers move seven figures on large estates.

Landing It Well

Treat the service as a first class tenant of the landing zone: delegated subnets planned in the hub and spoke address space with room for growth, the application spokes peered on the shortest path, DR designed with Data Guard across zones or to a second region’s deployment and rehearsed with the DR posts’ discipline, backups on Oracle’s tooling but restore drills on your calendar, and the seam instrumented, latency between app tier and database as a standing SLO, because the multicloud boundary is exactly where finger pointing lives during incidents and only your own measurements arbitrate. Migration onto it from on premises is Data Guard or RMAN based, the smoothest path in the whole migration catalog since both ends are Oracle. The strategic frame: this offering removes the infrastructure excuse from the Oracle conversation, leaving only the real question, which workloads are Oracle by requirement and which by inertia, and that question now gets answered per workload with a credible landing spot for both answers.

Cheers
Osama

The offering places OCI managed infrastructure in Azure regions: Exadata Database Service on dedicated infrastructure for the full control RAC estate, and Autonomous Database for the workloads that fit its managed model, both provisioned through Azure portal and APIs, billed through your Azure agreement (counting toward Azure commitments, which the FinOps posts recognize as real money), and reachable from your VNets through delegated subnets with single digit millisecond latency to your application tiers. The operational split to internalize: Oracle runs the database infrastructure with OCI tooling and SLAs, your Azure estate provides the applications, networking, identity federation, and monitoring integration, and the multicloud seam, two operational models touching, is where the design attention goes. Entra integrates for identity, diagnostics and metrics flow toward your Log Analytics through the integration paths, and the network design keeps database client traffic inside the region’s fabric rather than hairpinning anywhere.

The Decision Against the Alternatives

The sorting logic, workload by workload, extends the migration posts’ dispositions. Oracle Database@Azure wins where the Oracle features are the point: RAC availability, Data Guard topologies, Exadata performance characteristics the workload measurably uses, and the vendor certified stacks (the ERP and banking cores) whose support matrices name Exadata. Migration to PostgreSQL (the September paths) wins where the database is Oracle by history rather than dependency, the license and lock in savings funding the conversion, and the fraction of estates in that category is larger than DBA folklore admits, which the assessment tooling quantifies per schema. Oracle on Azure VMs remains the niche middle for modest databases needing OS control without Exadata, carrying the licensing care it always did. And staying on premises stops being defensible once the same Exadata exists in region, the datacenter exit case from this week loses its last anchor tenant. Licensing threads through every branch: BYOL versus license included, the support rewards programs, and the commitment mathematics deserve a spreadsheet with finance in the room, because the numbers move seven figures on large estates.

Landing It Well

Treat the service as a first class tenant of the landing zone: delegated subnets planned in the hub and spoke address space with room for growth, the application spokes peered on the shortest path, DR designed with Data Guard across zones or to a second region’s deployment and rehearsed with the DR posts’ discipline, backups on Oracle’s tooling but restore drills on your calendar, and the seam instrumented, latency between app tier and database as a standing SLO, because the multicloud boundary is exactly where finger pointing lives during incidents and only your own measurements arbitrate. Migration onto it from on premises is Data Guard or RMAN based, the smoothest path in the whole migration catalog since both ends are Oracle. The strategic frame: this offering removes the infrastructure excuse from the Oracle conversation, leaving only the real question, which workloads are Oracle by requirement and which by inertia, and that question now gets answered per workload with a credible landing spot for both answers.

Cheers
Osama