Working Principle

Basic Concepts

Table 1-2 lists the basic concepts related to IP address failover.

Relationship Among Logical Ports, Ethernet Ports, VLANs, and Bond Ports

Services are running on logical ports. Logical ports can be created based on Ethernet ports, VLANs, or bond ports. Figure 1-1 shows the relationship among logical ports, Ethernet ports, VLANs, and bond ports.

Figure 1-1 Relationship among logical ports, Ethernet ports, VLANs, and bond ports

Table 1-3 explains the meaning of each mark number in the preceding figure.

Working Principle

An IP address fails over from a faulty port to another port. In this way, services are switched from the faulty port to the available port, ensuring service continuity and improving reliability of paths for accessing file systems. This process is transparent to users. The essence is a service switchover between ports. The ports can be Ethernet ports, bond ports, or ports on VLANs.

• Ethernet port-based IP address failover application

  To improve reliability of paths for accessing file systems, you can create logical ports based on Ethernet ports. When the Ethernet port that corresponds to a logical port fails, the system will locate an available Ethernet port of the same type, delete the logical port from the faulty Ethernet port, and create the same logical port on the available Ethernet port to carry services, ensuring service continuity. Figure 1-2 shows Ethernet port-based IP address failover application.

  Figure 1-2 Ethernet port-based IP address failover application
  

  Host services are running on logical port A of Ethernet port A. The corresponding IP address is a. If IP address failover is enabled, once Ethernet port A fails, the storage system will automatically locate an available Ethernet port B, delete the configuration of logical port A on Ethernet port A, and create and configure logical port A on Ethernet port B. In this way, host services are quickly switched to logical port A on Ethernet port B. The service switchover is executed quickly and is transparent to users.

  • For V300R006C00, if all ports in a failover group are faulty and then a standby port recovers, you must run the change logical_port general command to set this standby port as the logical port's home port to recover services. For details about how to use the command, see the Command Reference of your product.
  • For V300R006C10 and later versions, if all ports in a failover group are faulty and then a standby port recovers, the system will automatically set this standby port as the logical port's home port and recover services.

• Bond port-based IP address failover application

  To improve reliability of paths for accessing file systems, you can bond multiple Ethernet ports to form a bond port. When an Ethernet port that is used to create the bond port fails, services are still running on the bond port. IP addresses fail over only when all Ethernet ports that are used to create the bond port fail. The storage system will locate available ports, delete logical ports on the faulty ports, and create the same logical ports on the available ports to switch services to the available ports. After the faulty ports recover, they take over their services again. Figure 1-3 shows bond port-based IP address failover application.

  Figure 1-3 Bond port-based IP address failover application
  

  Multiple Ethernet ports are bonded to form bond port A. Logical port A that is created on bond port A can provide high-speed data transfer. If both Ethernet ports A and B fail, the storage system will automatically locate bond port B, delete logical port A, and create the same logical port A on bond port B. In this way, services are switched from bond port A to bond port B. After Ethernet ports A and B recover, services will be switched back to bond port A. The service switchover is executed quickly and is transparent to users.

• VLAN-based IP address failover application

  You can create VLANs to isolate different services. When an Ethernet port on a VLAN fails, the storage system will automatically locate an available port of the same type, delete logical ports from the faulty port, and create the same logical ports on the available port. Then, services are switched from the faulty port to the available port.

  To implement a service switchover, you must create VLANs, allocate a unique ID to each VLAN, and use the VLANs to isolate different services. When an Ethernet port on a VLAN fails, the storage system will automatically locate an available Ethernet port with the same VLAN ID and switch services to the available Ethernet port. After the faulty port recovers, it takes over its services again. Figure 1-4 shows VLAN-based IP address failover application.

  Figure 1-4 VLAN-based IP address failover application
  

  Ethernet ports and their corresponding switch ports are divided into multiple VLANs, and different IDs are allocated to the VLANs. The VLANs are used to isolated different services. VLAN A is created on Ethernet port A, and the VLAN ID is 1. Logical port A that is created based on VLAN A can be used to isolate services. If Ethernet port A fails, the storage system will automatically locate VLAN B and the port whose VLAN ID is 1, delete logical port A, and create the same logical port A based on VLAN B. In this way, services are switched to VLAN B. After Ethernet port A recovers, services will be switched back to VLAN A.

  An Ethernet port can belong to multiple VLANs. If the Ethernet port fails, all VLANs will fail. You must switch services to ports of other available VLANs. The service switchover is executed quickly and is transparent to users.

• IP address failover relationship

  Before getting to know the IP address failover relationship, you must understand failover groups. Failover groups include a default failover group, VLAN failover group and customized failover group. IP address failover enables IP addresses to fail over between interface modules that use the same protocol. IP address failover policies are related to the location of a selected port and the MTU value of the port. You are advised to select a symmetric network for IP address failover (controllers for IP address failover must use interface modules of the same type). The priority of ports on controllers for IP address failover is as follows: a port (same location) on the peer controller, another port on the peer controller, a port on the current controller.

  • Default failover group (failover group of Ethernet ports and bond ports): The storage system automatically adds all Ethernet ports and bond ports to the default failover group.
  • VLAN failover group: When VLANs are created, the storage system automatically adds VLANs with the same ID to a failover group, that is, each VLAN ID has a failover group.
  • Customized failover group: You can add the ports to which IP addresses can fail over to the customized group. The port failover policy in the customized group does not require a symmetric network. During the failover, the system selects the same type of ports based on the order that the ports are added to the customized failover group. If such ports are unavailable, the system selects available ports of other types based on the preceding order.

    IP addresses not only can fail over between logical ports that are created based on Ethernet ports, but can also fail over between logical ports that are created based on other ports. Table 1-4 shows the IP address failover relationship.

• If logical ports are created based on Ethernet ports, bond ports, or VLAN ports, these ports can only be used for file system storage services.
• If logical ports are created based on Ethernet ports, the storage system will locate an available port in the failover group (preference is given to Ethernet ports and then to bond ports), delete the configuration of the logical ports created on the faulty port, and create the same logical ports on the available port. In this way, host services are switched to the available port, ensuring service continuity.
• If logical ports are created based on bond ports, the storage system will locate an available port in the failover group (preference is given to bond ports and then to Ethernet ports), delete the configuration of the logical ports created on the faulty port, and create the same logical ports on the available port. In this way, host services are switched to the available port, ensuring service continuity.
• If logical ports are created based on ports of VLANs, the storage system will locate an available VLAN port in the failover group, delete the configuration of the logical ports created on the faulty port, and create the same logical ports on the available VLAN port. In this way, host services are switched to the available VLAN port, ensuring service continuity.
• Ensure that the switches between the storage systems and application servers can receive broadcast packets and update ARP entries. Otherwise, IP address failover may fail.

  Some switches on which gratuitous ARP is disabled or anti-MAC address attack is enabled may fail to receive broadcast packets and update ARP entries during IP address failover of the storage system, resulting in an IP address failover failure. For the functions that affect the reception of broadcast packets and update of ARP entries, see the product documentation of the switches.