General Metro-DR prerequisites

Prerequisites for Metro-DR between two OpenShift® Container Platform clusters.

Meet the following requirements:
Note: Metro-DR solution is only supported on Global Data Platform storage.
Setting up the tiebreaker
  • Install IBM Storage Fusion HCI System in two separate data centers within a metropolitan distance from each other. A third site must be identified to host the tiebreaker, and a VM is needed to host the tie breaker. For the actual procedure to setup, see Preparing the tiebreaker.
Setting up a network
  • Network connectivity must be available between all three sites for both the Red Hat® OpenShift network and the IBM Storage Fusion HCI System storage network. The IBM Storage Fusion HCI System storage network is responsible for the synchronous data replication between both sites, and a high-bandwidth L3 connection must be used that has less than 40 milliseconds of latency and within 150 km. It is recommended that all switches used to connect the two clusters support jumbo frames as it improves performance.
  • The IBM Storage Scale core pod that runs on every IBM Storage Fusion HCI System node at both sites is assigned an IP address on the storage network. As the number of nodes in a cluster can grow over time, a default IP address range is automatically generated for each cluster. This default IP address range is displayed during the installation GUI for each cluster. If needed, you can provide a different range of IP addresses for the storage network. The IP address range must be different for each site and must be reachable across sites.
  • By default, IBM Storage Fusion HCI System reserves the 3201 VLAN ID for the storage network. If you need to customize the VLAN ID, record the customization in the planning worksheet. IBM SSR needs the completed worksheet to do the initial configuration of the appliance.
  • For more information about how to connect both the sites, see Metro-DR (Disaster Recovery).
  • Jumbo frames
    • Configure all the network devices through which the traffic flows from one site to another with Maximum Transmission Unit (MTU) size 9216 or higher. If your network does not support jumbo frames all the way from site 1 to site 2, then do the following steps:
      • Do not enable Jumbo frames during the installation of site 2.
      • If you configured site 1 with jumbo frames, reconfigure site 1 to disable it.
      For steps to enable or disable, see Configure GDP storage topic.
    • If you enable jumbo frames on the site 1, then you must enable it on the site 2 and on all network devices in between.
    • Both the client and IBM Storage Fusion HCI System switches must have matching packet sizes.
    • For optimal performance, IBM recommends that you enable the jumbo frames.
Setting up communication between the clusters
The Metro-DR solution involves network traffic between two IBM Storage Fusion HCI System clusters and the tie breaker. Update the firewall for the three sites to allow traffic between each of the three endpoints in the solution.
  • Site one and site two must allow communication between the IP addresses of the two IBM Storage Fusion HCI System clusters on both Red Hat OpenShift network and the IBM Storage Fusion HCI System storage network. On the Red Hat OpenShift, the network communications across sites take place over UDP port 4500. On the storage network, communication across sites takes place over TCP ports 1191 and 12345.
  • Site one and the tie breaker site must allow communication between the IP address on the storage network of site one cluster and the IP address of the tie breaker virtual machine. This communication takes place over 1191 and 12345.
  • Site two and the tie breaker site must allow communication between the IP address on the storage network of the site two clusters and the IP address of the tie breaker virtual machine. This communication takes place over 1191 and 12345.
  • For Metro-DR allowed ports, see Firewall requirements for IBM Storage Fusion HCI System.
Storage network
  • A default IP range is provided out of the box for the storage network.
  • If needed, you can provide a different range of IP addresses for the storage network; however, it must be different across sites.
  • Storage VLANs and storage networks are customizable for a single rack and must use only a private network for the storage network.
Admin network
Each cluster has a pod network that needs to communicate with each other in a Metro-DR setup. To achieve this communication, make sure that the pod network is different across the sites. By default, different values are available for site 1 and site 2 during installation, but you can override them.