Overview
Explore the power of query languages for distributed systems programming in this thought-provoking conference talk from Strange Loop. Delve into the semantics of query languages and their potential to simplify complex distributed systems programming. Discover Dedalus, a query language designed for distributed programming that abstracts away traditional programming details and prioritizes data and time. Learn how this approach enables programmers to reason about distributed correctness properties and lays the foundation for advanced program analysis tools. Follow Peter Alvaro's journey through the evolution of data-centric programming languages, from Dedalus to Bloom, Edelweiss, and Eve. Gain insights into the speaker's research on using data-centric languages to build scalable, predictable, and robust distributed systems. Examine topics such as abstraction, back-of-envelope calculations, stateless and stateful protocols, distributed language requirements, and the CALM Theorem. Uncover how Dedalus's unique design choices and logic programming tricks from model theory contribute to its effectiveness in distributed systems programming.
Syllabus
Intro
Abstraction: the good
Abstraction: the bad
Abstraction: the ugly
Back-of-envelope calculations
Hellerstein's inequality
Semantics: a richer example
Lineage and derivation
(stateless) protocols
(stateful) protocols
Distributed language requirements
A relativist model of time
Dedalus: Syntax
Syntax Sugar
Dedalus: stateful patterns
Ultimate Models
CALM Theorem
Dedalus: design choices
Logic programming tricks From model theory
Dedalus answers
Taught by
Strange Loop Conference