High Availability

This section describes how Canton can be run with high availability (HA). Support for HA is being added to the Canton components gradually. The specifics and configuration of HA for each component will be filled in as the implementation is completed. Furthermore, we are mostly starting with cold and warm standby solutions, and will gradually move to hot standby and active-active solutions to improve the mean time to recovery.

• Canton High Availability: Overview and Principles
• Replicated Participant Node Architecture
  • High-Level System Design
    • Why a Shared Database?
  • Participant Node Replica Monitoring and Fail-Over
    • Leader Election through Exclusive Lock Acquisition
    • Lock ID Allocation
  • Enforce a Passive Replica
    • Lock Loss and Fail-Over
  • Ledger API Client Fail-Over via Load Balancer
• Domain HA
  • Sequencer HA
    • Database Sequencer HA
  • Mediator and Domain Manager HA