Two controllers share drives; if one fails, the other takes over. Cost-efficient at small scale but performance drops ~50% if a controller fails.