Skip to main content

2024-03-15

· 2 min read

Present​

  • Got results from "split-brain" experiment with 100 nodes that basically confirm the faithfulness of the IOSim implementation and highlights some "interesting" consequences of Peras, namely that when the split-brain heals, it's not the longest chain that wins but the one with the most votes
  • Started work on implementing IOSim node that represents network interactions more faithfully, eg. using a pull model that mirrors the actual ChainSync protocol implemented in the cardano-node
  • Rust node passes the NetworkModel tests, with multiple nodes connected through a netsim-based layer. Aligning the implementations posed some interesting challenges, especially in the domain of time management (real vs. simulated time) and the scheduling of the various "actors" in the system
  • Research shared with engineering a new version of the protocol which introduces Certificates to aggregate votes in the block headers and mostly get rid of the problematic dangling votes concept
  • First meeting with PNSol on ΔQSD modelling of the protocol
  • Meeting with Spyros Voulgaris on Peernet on the topic of network simulator. This is the tool that's used by researchers when they need some numbers.
  • Starting significant redesign of IOSim node to cope with changes in the protocol

Issues​

  • We need to improve communication between research and engineering, and how changes are shared and notified