Skip to content

FEDERATION_THREAT_MODEL §3.3.2: is BFT ⌈2N/3⌉ the right DURABILITY ratio for verify-coord replication? (regions are replicas, not voters) #196

Description

@emooreatx

Spun out of CIRISPersist#408 (the infra-does-not-vote pin, now durable in persist's verify_coord.rs module docs per #407). The open threat-model question: QUORUM_WRITE_THRESHOLD = ⌈2N/3⌉ is a PBFT quorum for Byzantine voters, but regions carry no agency — a Byzantine region cannot forge a signed revocation, only withhold one (an availability/liveness fault orthogonal to the signature layer). If regions are cooperating replicas rather than mutually-adversarial authorities, a plain replication majority may be the honest durability model and ⌈2N/3⌉ is over-specified. Decision wanted in FEDERATION_THREAT_MODEL §3.3.2 before CIRISPersist#143 wires enforcement (which must enforce replication durability, never a vote, at whatever ratio this decides). Refs: CIRISPersist#408, CIRISPersist#407, CIRISPersist#143, CIRISVerify#77.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions