The talk links firmware, clusters, and alternative operating systems.
Key idea & notes ▾
WHY THIS MATTERSUnderstanding all three layers explains how real machines start and scale.
SOURCE Watch the video
Firmware is the first code a computer executes at power-on.
Key idea & notes ▾
WHY THIS MATTERSNothing else on a computer works until firmware finishes its job.
SOURCE Watch the video
Legacy firmware is proprietary, slow, and opaque.
Key idea & notes ▾
WHY THIS MATTERSOpaque firmware means you cannot verify what runs before your OS.
SOURCE Watch the video
Coreboot is minimal open-source firmware.
Key idea & notes ▾
WHY THIS MATTERSOpen firmware lets you audit and trust the earliest code on your machine.
SOURCE Watch the video
Coreboot separates hardware init from a swappable payload.
Key idea & notes ▾
WHY THIS MATTERSThis split lets one firmware base support many boot targets.
SOURCE Watch the video
Minimal auditable firmware shrinks the attack surface.
Key idea & notes ▾
WHY THIS MATTERSAttacks that run before the OS are the hardest to detect and remove.
SOURCE Watch the video
Clusters need thousands of machines to boot identically and reliably.
Key idea & notes ▾
WHY THIS MATTERSA boot flaw that is harmless once becomes catastrophic across ten thousand nodes.
SOURCE Watch the video
Infrastructure automates the full lifecycle of many machines.
Key idea & notes ▾
WHY THIS MATTERSManual management does not scale to thousands of servers.
SOURCE Watch the video
Open firmware makes fleet-wide boot control possible.
Key idea & notes ▾
WHY THIS MATTERSBoot policy you can script is boot policy you can automate and audit.
SOURCE Watch the video
A known-good boot state makes machines predictable.
Key idea & notes ▾
WHY THIS MATTERSYou cannot reliably debug or automate machines that start from unknown states.
SOURCE Watch the video
Plan 9 deliberately redesigns Unix's foundations.
Key idea & notes ▾
WHY THIS MATTERSStudying a clean redesign reveals which OS ideas were essential and which were accidental.
SOURCE Watch the video
In Plan 9 almost everything is a file.
Key idea & notes ▾
WHY THIS MATTERSA single uniform interface makes systems simpler to learn and compose.
SOURCE Watch the video
Each process gets its own customizable namespace.
Key idea & notes ▾
WHY THIS MATTERSPer-process namespaces enable clean isolation and flexible composition of resources.
SOURCE Watch the video
Plan 9 makes remote resources look and act local.
Key idea & notes ▾
WHY THIS MATTERSNetwork-transparent files make distributed systems far simpler to build.
SOURCE Watch the video
Plan 9's file model fits distributed cluster computing well.
Key idea & notes ▾
WHY THIS MATTERSA consistent model across machines reduces the complexity of managing clusters.
SOURCE Watch the video
Firmware, clusters, and OS design meet in real deployments.
Key idea & notes ▾
WHY THIS MATTERSHard-won deployment experience is what ties these abstract ideas together.
SOURCE Watch the video
Research OS ideas influence mainstream tools despite low adoption.
Key idea & notes ▾
WHY THIS MATTERSStudying niche systems reveals ideas that later become industry standard.
SOURCE Watch the video
Rethinking foundational layers exposes hidden assumptions.
Key idea & notes ▾
WHY THIS MATTERSThe biggest improvements often hide in layers no one thinks to question.
SOURCE Watch the video
Hands-on entry points make these topics approachable.
Key idea & notes ▾
WHY THIS MATTERSExperimenting cheaply and safely is the fastest way to understand systems.
SOURCE Watch the video
The three topics are layers of one control problem.
Key idea & notes ▾
WHY THIS MATTERSSeeing the whole stack as one story helps you reason about any system.
SOURCE Watch the video