Q: Multicluster Replication (bidirectional)

[Moezenka]

Hey,

I’m currently exploring options to set up a geo-redundant S3 storage using NooBaa.

The plan is to setup unidirectional replication both ways so that both buckets have the same data and have a load balancer distributing traffic to the S3 endpoints in an active/passive configuration.

          [Load Balancer]
              /     \
        (ACTIVE)   (PASSIVE)
        bucket-a   bucket-b

bucket-a resides in site-a.
bucket-b resides in site-b.

I need the bucket to be available in case of an outage.

While researching, I came across this Red Hat blog Multicluster replication in Red Hat OpenShift Data Foundation S3, which uses 4x buckets (2x per cluster).

Is it possible to set this up using only 2x buckets?
What is considered best practice for a setup like this?