In most cases where you do have more than one source (say datacenter for redundancy) you don't just tie them together directly, in simplest config you'd just switchovered from one to another if one failed.