This is Instant Connect’s preferred approach to providing data center level resiliency.

This approach, which is custom to Instant Connect, uses two independent Kubernetes clusters—be they single-node or multi-node—which are installed in different locations. Instant Connect runs on both clusters and replicates data and signaling messages between them to make the system appear as a single, logical entity. Because this solution involves application-layer data replication, network bandwidth and latency requirements are substantially relaxed; the two clusters can function (albeit in a 'split brain' mode) for hours, days or even weeks without permanent impairment.