Suppose a table is replicated at every site. An application that updates the table must update every replica. If one or more sites has failed, the update cannot be completed, so the application may be...

Suppose a table is replicated at every site. An application that updates the table must update every replica. If one or more sites has failed, the update cannot be completed, so the application may be unable to proceed until the site(s) come back online. How might this difficulty be avoided? Is there a disadvantage to your solution?