Disaster recovery topologies grow by accretion. A single standby becomes a standby plus a remote copy, then a remote copy plus a cloud target, then a development restore that doubles as a tertiary fallback. Each addition is justified on resilience grounds, and most are added without anyone mapping the new node against the licensing benefit that covered the first one. Software assurance grants one passive failover replica per licensed primary. Every copy beyond that single replica is a separate licensable instance unless it meets a narrow disaster recovery exception, and most multi site DR estates carry two or three uncounted instances by the time anyone looks. The exposure is real, but so is the recovery, because the same review that finds the uncounted nodes usually finds licenses wasted on replicas that qualified for the free benefit all along.
The free failover instance under software assurance covers exactly one passive secondary per licensed primary. A disaster recovery design with multiple standby copies has to account for every node beyond that one, because the benefit does not stretch to cover a second or third replica simply because it exists for resilience.
One passive secondary per licensed primary qualifies for the free benefit, provided the primary carries active software assurance and the replica stays genuinely passive. This is the node most teams correctly understand as free. The misunderstanding starts at the second copy.
A second standby for additional resilience, a geographically separate copy, or a cloud based replica added to the topology is a distinct instance. Unless it qualifies under a specific DR exception, it requires its own license whether it ever serves a query or not.
Disaster recovery exposure rarely comes from a single bad decision. It comes from three topology choices that each seemed prudent and that each added a licensable instance no one tracked.
An Always On availability group with a primary and two secondaries is a common high resilience design. The benefit covers one secondary. The second secondary is a full licensable instance, and most estates that built three node groups for safety never licensed the third node because the architecture predated anyone reading the failover terms closely.
Replicating an on premises primary to a cloud standby for site resilience adds a node that needs licensing, and the cloud platform changes how that license is provided. Whether license mobility, the failover benefit, or pay as you go covers the cloud copy depends on the configuration, and the default assumption that it is simply free is usually wrong.
A standby kept warm for failover that also runs reporting, test refreshes, or backups has lost the passive condition and become active. It now requires a full license even as the single free benefit, and the team is often unaware the resilience node has quietly turned into a working instance.
A correct disaster recovery position counts every node, applies the free benefit where it genuinely lands, and licenses the rest deliberately rather than discovering them in an audit. The exercise usually recovers spend as well as closing exposure, because over licensed and uncounted nodes tend to coexist in the same estate.
Every replica in the DR topology is mapped against the single free benefit and the DR exception. The nodes that genuinely qualify as the one passive failover instance stay free. The rest are licensed on your terms, often using software assurance license mobility or reserved capacity in the cloud to do it at the lowest defensible cost. The point is to surface every node before Microsoft does, so the conversation is about correct licensing rather than a settlement on instances you never knew were exposed.
The mirror image of the exposure is the standby that was fully licensed defensively when it qualified for the free benefit all along. Reclassifying those replicas as the free failover instance releases licenses that can cover the genuinely licensable DR nodes or defer purchases at renewal. In multi site estates built before the failover terms were understood, this recovery frequently offsets a meaningful share of the cost of licensing the nodes that were truly exposed.
The single free failover benefit, the DR exception that covers a true recovery copy, and the node by node review that licenses a multi site DR estate at the lowest defensible cost. Sent on request.
We count every node in the disaster recovery estate, apply the single free failover benefit where it genuinely lands, license the remaining copies at the lowest defensible cost, and reclaim the replicas that were over licensed when they qualified for the benefit. The DR line becomes a known number rather than an audit risk.