What Serverless Computing Is and Should Become

Let us examine an illustrative example from big data processing. Consider a simple query that might arise in an ecommerce setting: computing an average over 10 billion records using weights derived from one million categories. This workload has the potential for a lot of parallelism, so it benefits from the serverless illusion of infinite resources. We present two application-specific serverless offerings that cater to this example and illustrate how the category affords multiple approaches. One could use the AWS Athena big data query engine, a tool programmed using SQL (Structured Query Language), to execute queries against data in object storage.