Metro-DR (Disaster Recovery)

Metro-DR (Disaster Recovery) provides two-way synchronous data replication between IBM Storage Fusion HCI System clusters installed at two sites. When site disaster occurs, applications can be failed over to the second site. Because replication is synchronous, the Metro-DR solution is only available for metropolitan distance data centers with 40-millisecond latency or less.

You can deploy IBM Storage Fusion HCI System in data center site 1 and data center site 2, resulting in Red Hat® OpenShift® Container Platform clusters that run at each site. Two clusters are configured such that they share a synchronously replicated storage layer, allowing for data mirroring with zero Recovery Point Objective (RPO) between the two clusters. Both the Red Hat OpenShift clusters can host active workloads at the same time rather than keeping one cluster to be a cold standby. The IBM Storage Fusion HCI System storage network is stretched between both Red Hat OpenShift clusters so that data written to one cluster is automatically mirrored to the second cluster. A special tie breaker node is hosted at a third site and is used to determine which cluster is in charge of the data when the network between the two clusters is severed. Configuring a Metro-DR topology requires several network connections to be made between the two clusters and the tie breaker.

Note: In Metro-DR, the terminology scale admin network refers to the OpenShift Container Platform pod network, and scale daemon network refers to the IBM Storage Fusion HCI System storage network.

The following picture depicts the connectivity between Site 1 and Site 2.

Metro-DR solution depicting the connectivity between site 1 and site 2

The Metro-DR setup includes two sets of networks, namely internal storage network and Red Hat OpenShift network. In Metro-DR, the internal storage network of one site must connect to the other site and tiebreaker, that is, externalize your storage network. Similarly, the Red Hat OpenShift network of one site needs to connect to the other site.

OpenShift management network: The Red Hat OpenShift management network provides communication between the two OpenShift clusters, and is used by IBM Storage Fusion HCI System to coordinate replication activities, such as determining which applications are mirrored or triggering application failover. Every node in both clusters has an IP address in this network. The Red Hat OpenShift network across the sites must not be connected over Network address translation (NAT). You can decide to have the Red Hat OpenShift network across the sites in the same or different subnet.

IBM Storage Fusion storage network: The IBM Storage Fusion HCI System storage network is used for storage communication between nodes. In a Metro-DR configuration, the storage network must be connected between the two clusters to allow data replication. The storage network must also be connected to a tie breaker and the tie-breaker needs connectivity only on the storage network. This allows the tie breaker to choose which cluster is in charge of the data when the two clusters cannot talk to each other. The IBM Storage Fusion HCI System Storage network uses the 100G interface of each node and requires a high-bandwidth connection across the sites. The IBM Storage Fusion HCI System Storage network between the sites must not be connected through Network Address Translation (NAT). The IBM Storage Fusion HCI System storage network for each site must be in a different subnet than the other site and the tiebreaker.