Azure Arc extends the Azure control plane to SQL Server instances running anywhere: on premises, in another cloud, at the edge. For licensing, that changes the game in two directions. It gives you a continuous, accurate inventory of every instance and edition, which is the single most valuable input to any compliance position. And it opens pay as you go licensing and streamlined Extended Security Update delivery for estates that would otherwise be locked into a fixed license count. The catch is that Arc surfaces everything, including the instances you did not know were running an edition you are not licensed for. That visibility is an asset in a negotiation and a liability in an unmanaged estate.
The largest hidden risk in any SQL Server estate is the instance no one is tracking: the edition installed for a project years ago, the developer copy promoted to production, the instance inherited through an acquisition. Arc connects each instance to Azure and reports its edition, version, and core count continuously. That inventory is the foundation of every downstream decision, because both the cost optimization and the compliance defense start from knowing exactly what is deployed.
Arc assembles the same instance level inventory a Microsoft audit would seek to establish, except you hold it first. That ordering is the entire advantage.
Arc reports Developer edition instances quietly running production workloads, Enterprise instances where Standard would do, and instances with no license assigned at all. In a managed estate this is a gift, surfacing the exposure before Microsoft does. In an unmanaged one it is a finding waiting to be made, which is why connecting Arc without a plan to act on what it reveals is the wrong sequence.
Arc enables pay as you go SQL Server licensing on instances anywhere, billed through the Azure subscription by the core hour rather than as a fixed perpetual purchase. For the right workloads this is a structural cost improvement. For the wrong ones it is more expensive than the license you already own. Knowing which is which is the whole exercise.
Pay as you go fits instances that do not run continuously or whose future is uncertain. A workload spun up for a project, a seasonal instance, a database awaiting migration or decommission, an environment in flux after an acquisition: paying by the hour avoids buying a perpetual license for capacity that will not persist. It also converts a capital purchase into operating spend that flows through the Azure commitment, which can be advantageous for both the budget treatment and the consumption math against a MACC.
For an instance that runs continuously for years, perpetual licensing with software assurance is almost always cheaper than paying by the hour indefinitely, because the hourly rate is priced to favor Microsoft over a long horizon. The error is moving a stable production estate to pay as you go for the convenience and paying a premium for the privilege. We model the breakeven for each instance and keep the steady workloads on owned licenses while routing only the variable ones to consumption.
For SQL Server versions past end of support, Extended Security Updates keep the instance patched. Arc has become the streamlined mechanism for enrolling and delivering ESUs to instances running outside Azure, which matters for any estate still operating older versions it cannot immediately migrate.
Connecting an out of support instance through Arc enables ESU enrollment and automatic patch delivery without the manual key management the older ESU process required. The instance stays supported while the migration plan plays out, rather than running unpatched.
ESUs are a bridge to a migration, priced to encourage you to move. Carrying an instance on ESUs indefinitely is expensive by design. The right use is buying the time to migrate or modernize, with a dated plan to exit, not treating perpetual ESU as a steady state.
SQL Server instances migrated to Azure receive Extended Security Updates at no additional charge. That asymmetry is deliberate and is a genuine factor in the migrate versus stay decision, but it should inform the decision rather than dictate it, weighed against the full cost of the move.
The inventory model Arc provides, the pay as you go breakeven analysis by workload type, the ESU enrollment path, and the visibility risks to manage before connecting an unprepared estate. Sent on request.
We assess the SQL Server estate through the lens Arc provides, model where pay as you go beats perpetual licensing and where it does not, plan the ESU bridge for out of support instances, and manage the exposure that visibility surfaces so the inventory becomes your negotiating asset rather than the auditor's finding.