In a multi-node Kubernetes cluster, IBM API Connect's management subsystem handles stateful operations, including OAuth token management. To ensure high availability (HA) and data consistency for these operations, the underlying distributed database (e.g., PostgreSQL) relies on a consensus algorithm. The core principle of these algorithms is quorum, which is the minimum number of server nodes that must be active and agree on an operation for it to be committed. This prevents data inconsistencies ("split-brain") and ensures that critical OAuth operations are reliably processed even if some nodes fail. The requirement for a minimum of three management nodes in an HA API Connect deployment directly reflects this principle.

1. Official Vendor Documentation: IBM API Connect v10.0.x Knowledge Center

"Management subsystem requirements". The documentation specifies

"For a production installation

a 3-node cluster is recommended for high availability." This three-node minimum is a standard architecture for establishing a quorum-based consensus system for the underlying data store that manages OAuth and other configuration data.

Source: IBM. (2023). Management subsystem requirements. IBM API Connect 10.0.x Documentation. Retrieved from www.ibm.com/docs/en/api-connect/10.0.x?topic=kubernetes-management-subsystem-requirements.

2. University Courseware/Academic Publication: The concept of quorum is fundamental to consensus algorithms like Raft

which are used in modern distributed systems. A leader can only append a new entry to the log once a majority of servers (a quorum) have responded

ensuring consistency.

Source: Ongaro

D.

& Ousterhout

J. (2014). In Search of an Understandable Consensus Algorithm. Proceedings of the 2014 USENIX Annual Technical Conference (USENIX ATC '14). Stanford University. (Section 3.2

"Leader election

" and Section 3.3

"Log replication

" describe the necessity of a majority/quorum for operations).

3. Official Vendor Documentation (Kubernetes): The official Kubernetes documentation on running stateful applications explains the challenges of managing distributed data stores. These systems

which would manage OAuth state

inherently rely on consensus and quorum to maintain a consistent state across replicas.

Source: Kubernetes Authors. (2023). Running Stateful Applications. Kubernetes Documentation. Retrieved from kubernetes.io/docs/tutorials/stateful-application/running-stateful-application/. (This document discusses the principles of stateful sets and persistent storage

which are prerequisites for building HA data services that use quorum).