A reconcile operation is either manual or automatic.

After completing some changes to a version, an editor selects another version—the target version—against which to reconcile.

The relative cost of a reconcile operation is directly related to the complexity of the data model and the number of classes modified in the version being reconciled. For example, if geometric networks, topologies, or relationships are being reconciled, additional processing is required to ensure all associated geodatabase behavior is applied correctly to preserve the integrity of the data.

Although the volume of changes can also affect performance, as a general rule it?s more efficient to reconcile a large number of changes in batches rather than reconciling a small number of changes at frequent intervals. The main performance cost of the reconcile process is opening the datasets in each version during the reconcile; in most cases, there is less cost associated with moving the changes between versions. However, if the volume of changes to reconcile is considerable, it may be more efficient to reconcile the changes in smaller batches.