An Oracle to SQL Server migration is the second-most common Oracle Database displacement we cost on the buyer side, after PostgreSQL. The licensing arithmetic is materially different from a move to Postgres because SQL Server is a commercial product with its own metric, its own audit regime, and its own renewal economics. Treating SQL Server like "free Postgres but on Windows" is the single most common mistake we see in vendor-written migration proposals. This article walks the forensic licensing implications of an Oracle to SQL Server migration: edition mapping, two-core-pack arithmetic, Software Assurance economics, the Microsoft Enterprise Agreement lever Oracle's renewal team is paid to defend against, and the redline package we run against both vendors at the negotiation table. Real numbers, sourced from 600+ engagements and the $1.8B in Oracle spend we have advised.
The Oracle to SQL Server migration conversation has shifted twice in the last decade. It went quiet in 2017 to 2020 as PostgreSQL economics improved and AWS RDS for Oracle reduced the friction of staying on Oracle in a cloud-first estate. It came back in 2023 to 2026 for three reasons. First, Microsoft's licence mobility through Software Assurance now genuinely supports a managed-Azure-SQL-MI landing zone that is operationally close enough to Oracle EE for the bulk of transaction workloads. Second, Oracle's annual support uplift has compounded out of pocket — at 4 to 8 percent compounding, the 2018 support bill is now 50 percent larger than the original net licence, with no incremental functionality delivered. Third, Microsoft FastTrack now subsidises Oracle to SQL Server migrations directly through Azure Migrate credits when the destination is Azure SQL Managed Instance, which materially changes the year-one cash position.
None of these reasons are stand-alone sufficient. What makes an Oracle to SQL Server migration credible in 2026 is the combination — managed destination, vendor-funded migration, and a forensic buyer-side TCO model that compares like-for-like at the entitlement level, not at the marketing slide level. The licensing implications of an Oracle to SQL Server migration are what most teams understand least. The base licence comparison is straightforward; the option mapping is not; the Software Assurance economics are not; and the audit reconciliation of the residual Oracle estate during cutover is where most CFOs find a $400K to $2.1M back-licence claim sitting in their lap. Read the Oracle audit guide for the back-licence claim mechanics that the migration window creates.
The metric mapping is the first place a vendor TCO model lies. Oracle's Processor Metric counts cores then multiplies by the Core Factor Table value. For x86 processors (Intel Xeon, AMD EPYC) the factor is 0.5. So a 32-core Intel Xeon server consumes 16 Oracle Processor licences. SQL Server licences are sold in two-core packs, with no core factor. The same 32-core server consumes 16 SQL Server two-core packs — a one-to-one count match at the pack level, but the per-core list price is different.
At Oracle list of $47,500 per Processor licence (Enterprise Edition) and SQL Server Enterprise list of $13,748 per two-core pack, the like-for-like comparison on a 32-core Xeon server is:
| Line item | Oracle Enterprise Edition | SQL Server Enterprise |
|---|---|---|
| Cores | 32 (16 after 0.5 core factor) | 32 (no factor) |
| Licences | 16 Processor | 16 two-core packs |
| Per-licence list | $47,500 | $13,748 |
| Net licence list | $760,000 | $219,968 |
| Annual maintenance (Oracle 22% / SA ~25%) | $167,200 | $54,992 |
| 5-year licence + maintenance (no uplift) | $1,596,000 | $494,928 |
At base-edition only, SQL Server Enterprise is roughly 31 percent of the Oracle EE list cost for the same physical compute. The gap widens once Oracle's option stack is loaded — Partitioning, Diagnostics Pack, Tuning Pack, Advanced Compression, and Active Data Guard collectively add $52,500 to $74,000 per Oracle Processor at list, none of which have a SQL Server equivalent line item because the SQL Server equivalents are bundled inside Enterprise edition.
The edition mapping is the second place vendor TCO models simplify too far. Oracle's two production editions (Enterprise Edition and Standard Edition 2) do not map cleanly to SQL Server's two production editions (Enterprise and Standard). The differences matter for the buyer-side model because they determine whether the migration drops the workload into a cheaper SQL Server edition or pins it in Enterprise because of a single feature dependency.
The trap we see most often is teams pricing the migration as Oracle EE → SQL Server Standard because the surface analysis says the workload "could" run on Standard, then discovering during refactor that the partitioning behaviour or the in-memory column store dependency forces SQL Server Enterprise. The buyer-side migration TCO model has to score each workload at the feature level before committing the edition assumption. See our Oracle Database edition comparison for the upstream feature inventory the SQL Server edition decision depends on.
Independent, buyer-side. Edition mapping, core-pack arithmetic, Software Assurance modelling, and the dual-vendor redline package. Fixed-fee, ten-day turnaround.
The most under-modelled saving in an Oracle to SQL Server migration is the Oracle Options stack that maps to SQL Server Enterprise-included features. Oracle charges separately for Partitioning, Advanced Compression, Diagnostics Pack, Tuning Pack, Real Application Testing, Database In-Memory, Active Data Guard, and Multitenant. SQL Server includes nearly all the functional equivalents inside Enterprise edition. This is where the 4 to 6x list-comparison gap on a real Oracle Enterprise workload shows up.
| Oracle Option | Oracle list / processor | SQL Server Enterprise equivalent | SQL Server cost |
|---|---|---|---|
| Partitioning | $11,500 | Table partitioning | Included in Enterprise (and Standard since 2016) |
| Advanced Compression | $11,500 | Row / page / columnstore compression | Included in Enterprise |
| Diagnostics Pack | $7,500 | Query Store, Extended Events, DMVs | Included in all editions |
| Tuning Pack | $5,000 | Database Engine Tuning Advisor, Query Store hints | Included in all editions |
| Real Application Testing | $11,500 | Distributed Replay (deprecated 2022) | 3rd party (Qlik, Quest) or limited native |
| Database In-Memory | $23,000 | In-Memory OLTP, Columnstore | Included in Enterprise |
| Active Data Guard | $23,000 | AlwaysOn Availability Groups (readable secondary) | Enterprise + SA (passive node free) |
| Multitenant | $17,500 | Contained Databases / multiple DBs per instance | Included in all editions |
| Spatial and Graph | $17,500 | SQL Server spatial types + graph tables | Included in all editions |
The total Oracle option stack on a fully-optioned Enterprise Edition workload (Partitioning, Compression, Diagnostics, Tuning, In-Memory, ADG, Multitenant) sums to $103,500 per processor at list. SQL Server Enterprise includes every functional equivalent for $13,748 per two-core pack. This is the line item that produces the real arithmetic gap. Modelling the migration only against the base licence comparison misses 60 to 70 percent of the recurring saving. Read the Oracle database licensing guide for the full options taxonomy that has to come out of the Oracle side of the model.
Software Assurance is the Microsoft contractual benefit bundle that runs alongside SQL Server licences. It is sold under an Enterprise Agreement or a Server and Cloud Enrollment and costs roughly 25 percent of net licence per year. SA looks like Oracle's 22 percent annual support on paper. The economics are not the same. Oracle's 22 percent buys vendor patch support and version-upgrade rights. SA buys version-upgrade rights, License Mobility through SA, Failover Server rights, Step-Up rights, planning services, and TechNet benefits. The Failover Server right and the License Mobility right are the two SA benefits that change the migration math.
The Failover Server right under SA allows one passive instance per active licensed instance at no incremental cost, as long as the passive node only handles failover and not query traffic. This is the single most common saving in a like-for-like comparison against Oracle Active Data Guard, which costs $23,000 per processor for what is functionally a similar feature. License Mobility through SA allows the licence to move between on-premises servers, Azure VMs, and qualifying third-party clouds within 90 days, which removes the friction of cloud landing-zone changes during the migration.
Over a 5-year horizon SA is more expensive than Oracle's 22 percent line in absolute terms (25 percent versus 22 percent of a smaller base, but the SA covers a larger feature surface). Customers who terminate SA after year three to bank the saving lose the Failover Server right on every passive node, which usually means re-licensing the secondary. The model has to include the SA stay-or-go decision explicitly and not assume SA discontinues silently. The Microsoft vs Oracle renewal tactics piece covers how the SA renewal cycle plays into a dual-vendor negotiation.
The case is from a 2025 engagement. A US financial services firm running a customer ledger on Oracle Enterprise Edition with Partitioning, Advanced Compression, Diagnostics, Tuning and Active Data Guard. Twelve Processor licences on a 24-core Intel Xeon estate. Net Oracle licence stack $1.16M, annual support $255K, 7 percent annual uplift forecast. Renewal approaching with a 38 percent uplift on the table. The board approved a buyer-side TCO model comparing stay-on-Oracle, migrate-to-SQL-Server-on-Azure-VM, and migrate-to-Azure-SQL-MI.
| 5-year total (USD) | Stay on Oracle | SQL Server Enterprise on Azure VM | Azure SQL Managed Instance |
|---|---|---|---|
| Database licence (5-year) | $1,160,000 (sunk) | $164,976 (12 two-core packs) | $0 (PaaS, vCore pricing) |
| Maintenance / SA / vCore consumption | $1,495,000 (22% + uplift) | $206,200 (5yr SA) | $1,260,000 (vCore General Purpose) |
| Audit reserve (LMS during migration) | $0 modelled | $240,000 | $240,000 |
| Refactor (PL/SQL → T-SQL, SSMA + manual) | $0 | $1,420,000 | $1,420,000 |
| DBA labour delta (5-year) | $0 baseline | ($380,000) saving | ($620,000) saving |
| Dual-run cost | $0 | $430,000 | $430,000 |
| Microsoft FastTrack credit | $0 | ($180,000) credit | ($240,000) credit |
| 5-year total | $2,655,000 | $1,901,176 | $2,490,000 |
The interesting outcome on this engagement is that the buyer-side model showed Azure SQL MI as the cleaner operational target but SQL Server Enterprise on Azure VM as the cheaper five-year model. The client elected to use the model as renewal lever rather than execute. Oracle's deal desk authorised a 53 percent discount off renewal list with a 0 percent uplift cap for the term, on the back of the credible Microsoft EA + FastTrack threat. The migration paused; the renewal saving over the contract term was $1.41M against the would-be uplifted Oracle line. See the closely-related Fortune 500 bank EA restructure case study for an adjacent engagement pattern.
Negotiating Oracle and Microsoft simultaneously? We sit on the buyer side and run both renewals against each other. The lever is in the shape of the threat package.
The dual-vendor negotiation that an Oracle to SQL Server migration enables is the highest-value redline opportunity in enterprise software. Each vendor's account team knows the other is there. Each is paid to defend share. The buyer-side redline package below has produced 50 to 70 percent off list at Oracle renewal and 35 to 55 percent off list on the Microsoft EA, in the same calendar quarter, on engagements where the migration TCO was credible.
The buyer-side advisor's job is to keep both vendors in the room without letting either drive the framing. Oracle's renewal team will try to anchor the comparison on Oracle list versus SQL Server list and skip the option stack. Microsoft's account team will try to anchor on Azure SQL MI consumption rather than Server & Cloud Enrollment subscription. The forensic model defends against both anchoring moves. The Audit Defence service covers the LMS audit risk that opens up during the migration window — Oracle's playbook on RFP-driven displacement is well-documented and the audit reserve has to be sized into the model.
At list, Oracle Enterprise Edition is $47,500 per processor and SQL Server Enterprise is $13,748 per two-core pack (so $6,874 per core list). Once Oracle's 0.5 core factor for x86 and SQL Server's full core count are applied, the per-physical-core list comparison is roughly $11,875 Oracle EE versus $6,874 SQL Server. Add the Oracle options stack (Partitioning, Diagnostics, Tuning, Compression, ADG) and the gap on a typical Oracle Enterprise workload widens to 4 to 6x.
Yes. Software Assurance at roughly 25 percent of licence list adds version-upgrade rights, Failover Server rights, and License Mobility through SA. The Failover Server benefit alone eliminates a passive-server licence requirement that Oracle Active Data Guard charges $23,000 per processor for. Most migrations capture this saving but neglect to model the SA renewal cycle past year five.
Schema and data migration land in 8 to 14 weeks using SQL Server Migration Assistant on a workload under a terabyte. PL/SQL to T-SQL refactor is the long pole and runs 6 to 18 months depending on PL/SQL volume. Cutover including dual-run typically spans 12 to 24 months end-to-end. Estates with heavy use of Advanced Queueing or Oracle Spatial run longer.
Yes, when the migration package is credible. A costed migration plan with a Microsoft Enterprise Agreement, a named Microsoft FastTrack engagement, and a sequenced cutover schedule routes the Oracle renewal to the deal desk. Discount authority shifts from 20 to 30 percent to 50 to 70 percent. The migration does not have to run for the model to pay for itself many times over at renewal.
PL/SQL packages with session state and Advanced Queueing are the two most common blockers we see. T-SQL has no direct equivalent to PL/SQL packaged session variables, and Service Broker is a partial AQ replacement at best. Most blockers add 15 to 35 percent to the manual refactor budget, which is why the buyer-side TCO model has to score them line by line before signing the EA. The buyer-side Oracle to SQL Server migration licensing analysis ends where the refactor evidence begins.
Twice a month. Oracle pricing moves, audit-defence tactics, migration economics. Written by former Oracle insiders. Independent and buyer-side only.
No spam. Unsubscribe any time. Independent — not affiliated with Oracle Corporation.