Distributed Systems
Notes on correctness under partial failure: state machines, consistency, replication, and making invariants survive real deployments.
Best starting points
Termination Is a Security Boundary: HotStuff Under UC, Delay Attacks, and the Uncomfortable Gap to Rust
Paper note (March 2026): a UC-style termination proof for HotStuff, the real invariant it relies on, and what changes when you ship it as a low-level Rust system under adversarial latency.
Hybrid Schemes and Protocol Agility
Deep dive (April 2026): hybrid key establishment is a narrow hedge (HNDL), not “post-quantum TLS”. The hard part is suite identity, transcript binding, and AND-semantics for dual signatures.
Stateful Signatures Are a Distributed Systems Problem: XMSS/LMS Without Index Reuse
Deep dive (April 2026): stateful hash-based signatures look like “just PQC”, but one index reuse is a catastrophic key-management failure. Model the invariant, then build the allocator like a consensus component.
The Leaf Is the Hot Path: Signature Placement in Post-Quantum TLS (ML-DSA vs SLH-DSA)
Paper note (April 2026): experiments show SLH-DSA in the server leaf collapses TLS 1.3 handshakes by ~10^3×. PQ migration is a certificate-hierarchy and cost-concentration problem, not an algorithm swap.
Post-Quantum Readiness at the Edge: Constraints and Migration
Engineering notebook entry (December 2021): Post-Quantum Readiness at the Edge: Constraints and Migration.
Roadmap
- State machine + invariants first
- Failure modes + partitions as default
- Replication/consensus + membership changes
- Operational evidence: SLOs, rollbacks, and drills