A virtual storage volume that includes storage space on potentially many different physical disks may be mounted at a compute node. In order to replicate the virtual storage volume to a target compute node, a sequence of snapshots may be created that capture the state of the virtual storage volume at a sequence of points in time. The data associated with these snapshots may be transferred to the target compute node. When the difference between the snapshots is sufficiently small, the target virtual storage volume may be resynchronized.