Where all the ladders start

Professor Peter Alvaro from UCSC presents a reflective look at twenty years of distributed systems research, examining whether the field – and he himself – is making meaningful progress toward building correct and robust distributed systems.

He walks through four major research chapters: first, the development of monotonic programming languages like Bloom that eliminate coordination by design; second, lineage-driven fault injection (LDFI) as a testing methodology that uses databases to reason about distributed systems; third, the realization that redundancy isn’t always beneficial; and fourth, recent work on metastable failures and simulation techniques.

Throughout, he demonstrates how transplanting ideas across domains leads to progress, while acknowledging that despite significant advances, the field may never reach a unified solution and must instead rely on defense in depth approaches.