Metro disaster recovery

Metro DR is a variation of the regional DR topology and also based on 2 managed Red Hat OpenShift clusters and an RHACM deployment for orchestration.

The key differences are:

  • Low latency between the nodes is essential. Like a stretched cluster, the round-trip latency must be below 5 ms round-trip.
  • IBM Fusion Data Foundation must be deployed in external mode and uses externally deployed stand-alone Ceph storage cluster (IBM Storage Ceph), which is stretched (shared) across the 2 data centers. With the common storage, the asynchronous replication can be avoided and a Recovery Point Objective (RPO) of zero can be achieved.
  • The stretched Ceph cluster needs an arbiter node to resolve inconsistencies in case one of the Ceph OSDs fails.
  • Both Red Hat OpenShift clusters can be active.
Note:
  • The external Ceph cluster is based on IBM Storage Ceph and can be deployed only in an x86 environment, as Ceph is currently not supported as a stand-alone product on IBM Z.
  • Red Hat Ceph Storage is being rebranded to IBM Storage Ceph.
  • Metro DR is supported on IBM Z and IBM® LinuxONE. Metro DR is not supported on IBM zCX.
Figure 1. Metro DR setup
The content of this image is explained in the surrounding text.