A server arrives with the expected quantity of memory, but the operating system reports less capacity than the inventory suggests. Before ordering replacement DIMMs, establish which number is being compared: installed capacity, firmware-recognized capacity or memory made available to software.
Those figures serve different purposes. A useful investigation follows the gap through the system, rather than assuming the module inventory or the operating-system reading must be wrong.
Mirroring uses capacity for a second copy
Dell’s PowerEdge R940 documentation describes Mirror Mode as maintaining two copies of memory data, leaving half of installed physical capacity available. It also specifies matching-module and population requirements for that platform. This is a documented reliability tradeoff, not evidence that half the DIMMs have failed. Dell’s R940 memory-mode guidelines explain the implementation.
Two copies share the physical total
One copy of the memory data
Duplicate data for the supported reliability mode
Use the server’s own service guide to identify supported modes and layouts. Names that sound similar do not establish identical behavior across generations. For example, the same Dell guide describes rank sparing separately from mirroring; the two should not be treated as interchangeable capacity calculations.
Reconcile the totals in three steps
Start with the physical inventory. Record module capacities, quantities and occupied slots. Then compare that inventory with the firmware or management-controller view. If a populated slot is unrecognized or reports an error, retain the event logs and investigate the discrepancy using the vendor’s diagnostic procedure.
Next, record the active memory mode and the expected capacity under that mode. Finally, compare the result with the operating-system or hypervisor total. Keep screenshots or exported reports with the same observation date so that a later configuration change does not confuse the diagnosis.
A small discrepancy has other possible explanations. Dell separately documents memory used by UPI Prefetch and Sub NUMA Cluster settings on relevant PowerEdge systems. That example shows why a capacity difference alone is insufficient to diagnose a hardware fault. Dell’s installed-versus-available memory note describes those reservations.
Make a reliability decision before changing the mode
If mirroring explains the difference, the next question is operational: why was it enabled? Review the workload’s availability requirements, the original build specification and any service policy governing the machine. Recovering capacity changes the configuration’s protection strategy.
Have the application owner and infrastructure team agree on the intended outcome before scheduling a change. Record the previous settings, expected capacity, vendor prerequisites and rollback procedure. Validate the application after the maintenance window rather than declaring success solely because the memory total increased.
An upgrade proposal should show usable capacity under the mode the organization intends to keep. Quoting only the physical total can overstate the resources available for the next workload and leave the purchasing team comparing unlike systems.
Give buyers both the inventory and the configuration
For a complete server sale, describe installed RAM and the currently reported usable capacity separately, alongside the active memory mode. If selling the modules individually, keep their identity and condition records distinct from the configuration of the source server.
A clear handover avoids an unnecessary dispute: the buyer can see what hardware is included, what the current settings do and what must be validated before deployment.
Prepared November 10, 2026. Dell’s R940 manual supplies the mirroring example; its PowerEdge support note supplies an additional reservation example. Investigation and handover steps are Ocitech’s recommendations. This guide does not prescribe a memory mode for every platform.