Remote Copy over IP (RCIP) network transport requirements

The physical connections between all storage systems used with Remote Copy over IP (RCIP) are through an IP-capable network. Each link between a pair of storage systems is a logical link between a controller node on one storage system and a controller node on the other storage system in the configuration. These links use an Ethernet port from each of the nodes in the storage systems.

To use an IP network for Remote Copy, the following requirements must be met.
  • A minimum of two storage systems that are supported for Remote Copy over IP.

  • A minimum of two links from two different nodes for each storage system in the Remote Copy configuration. A maximum of one IP link per node.

    Utilizing like node pairs (such as 0/1, 2/3) on each storage system is highly recommended, but is not a requirement.

  • Different IP subnets for the RCIP Ethernet port and the management Ethernet port of the system controller node.

  • Unique IP addresses for all RCIP ports.

  • A firewall that allows traffic only over the designated RCIP ports, TCP 5785 and 5001.

  • The Remote Copy subnet can be shared with other I/O traffic, but a guaranteed amount of bandwidth must be dedicated to Remote Copy. The RCIP network must be sized appropriately to replicate the data being generated. Without a guaranteed amount of bandwidth dedicated to RCIP, the Recovery Point Objective (RPO) for the solution cannot be assured.

    Review the Service Level Agreement with the network provider to determine if the provider can temporarily reduce bandwidth without notice. Check if there is a minimum bandwidth.

    When replicating synchronously over RCIP, the bandwidth must be large enough to ensure that no performance issues arise from replicating data synchronously. For asynchronous periodic replication, the bandwidth must be large enough to ensure that the RPO can be met.