The document discusses different approaches that social media companies take to store and query massive amounts of social data in real-time. Twitter developed Earlybird to enable real-time search through fast indexing and concurrency management. Facebook built TAO to store its social graph data, aiming to provide very fast response times and high read availability at scale. LinkedIn created Voldemort, a highly scalable distributed key-value store, to meet its data partitioning needs through seamless replication across many nodes.