Joe Hellerstein, professor at UC Berkeley and co-founder of Trifacta, presented a keynote on serverless computing and distributed data at the 2020 Distributed SQL Summit. He explained that serverless computing has limitations, including large I/O bottlenecks, short lifetimes for functions without session context, and no inbound network communication allowed. To overcome these challenges, Joe discussed avoiding coordination in distributed systems, introducing the CALM theorem, which describes when exactly you can correctly avoid coordination. The theorem highlights the importance of monotonicity, a property that allows programs to produce more output as input grows. Currently, researchers at Berkeley are working on a multi-component stack called Hydro, which includes an autoscaling key value store and a stateful serverless platform called Cloudburst, designed to decouple compute and storage while maintaining consistency across caches. The goal is to build scalable systems that can run without coordination, enabling "crazy fast" parallelism and smooth scalability.