Adventures in Spacetime
summary
Argues that software is the slowest medium and models networks/software with a measurable refractive index (ideal time vs actual time), demonstrates CAP theorem as an impossibility (choose consistency or availability under partition), and shows practical fixes: use Lamport clocks / logical timestamps and accept partial failure. Cites real incidents (O2 outage from an expired certificate), physics limits (London–New York ~19 light‑ms, making 10 ms global sync impossible), and tools like NTP and causal ordering to design around limits.
AI-written summary · about
chapters (14)
- 0:00🚀 Introduction and Overview of the Talk
- 2:00What is Spacetime? The Fusion of Time and Space
- 5:00Understanding the Speed of Light and Its Role in Software Development
- 9:00The Impact of Refraction and the Refractive Index on Communication
- 12:00Software as the Slowest Medium: The Role of Software in Slowing Down Hardware
- 17:00Why Networks and Software Have a Refractive Index
- 20:00Moving Up the Layer Stack: From Network Layers to Microservices
- 25:00Introduction to Distributed Systems: Challenges and Assumptions
- 28:00The Fallacies of Distributed Computing
- 32:00CAP Theorem: The Impossibility of Achieving Consistency, Availability, and Partition Tolerance Simultaneously
- 36:00The Myth of Infinite Bandwidth and the Reality of Network Constraints
- 39:00Time and Causality: The Limitations of Distributed Systems
- 42:00Lamport Clocks and Logical Time: A Solution for Event Ordering
- 45:00Closing Thoughts on Relativity, Time, and Software Development
related talks
27mDistributed Systems, Like It or Not
35mNobody Knows What Time It Is
28mRock around the clock (synchronization
59mOn Interplanetary and Relativistic Distributed Computing
29mHorrors of Distributed Systems
43mWhat We Talk About When We Talk About Distributed Systems
50mDesigning Fault-Tolerant Software with Control System Transparency
32mWas Leslie Lamport Right?