Level-by-level, array-backed BFS is exactly what large-scale graph engines like Facebook's Unicorn search index and academic frameworks like GAP/Ligra use instead of pointer-chasing queues, because a compressed-sparse-row adjacency list turns random graph traversal into mostly-sequential scans. At social-network or web-crawl scale, this single change is routinely the difference between a traversal finishing in seconds versus minutes.