zEnterprise System and System z10 OSA-Express Customer's Guide and Reference
Previous topic | Next topic | Contents | Contact z/OS | Library | PDF


ARP Takeover

zEnterprise System and System z10 OSA-Express Customer's Guide and Reference
SA22-7935-15

ARP Takeover is designed to preserve Host to Client connectivity for both OSA and Host Operating System failures. When an Start of change OSA-Express feature End of change fails, the Host Operating System will move the IP Addresses from the failed OSA to the backup OSA. When a Host Operating System fails, a backup Host Operating System will register the IP Addresses from the failing Host to OSA.

When TCP/IP is started in QDIO mode, it downloads all the IP addresses it needs to use for routing and stores them in each OSA-Express feature. This service of QDIO architecture occurs automatically for OSD channels.

For OSA-Express features set up as Start of change CHPID types End of change OSE (non-QDIO), you must define multiple IP addresses in the OAT using OSA/SF Start of change (or OSA/SF on HMC exclusively for the OSA-Express5S and OSA-Express4S features) End of change as all the ARP functions reside in the Host TCP/IP Stack. The OSA-Express then responds to ARP requests for its own IP address, as well as for virtual IP addresses (VIPAs). The OSA that is planned to takeover the ARP responsibility should have its own IP address, VIPA address and the IP addresses that it is planned to takeover in its OSA Address Table.

If an OSA-Express port fails while there is a backup OSA-Express port available on the same network or subnetwork, TCP/IP informs the backup OSA which IP addresses (real and VIPA) to take over, and the network connection is maintained.

Figure 1 shows two OSA-Express features (each having just 1 port per CHPID and 1 CHPID per OSA), each storing the same four IP addresses that were either downloaded by TCP/IP for OSD channels, or defined in the OAT for OSE channels. Figure 1 shows how this scenario works:

Figure 1. IP Address Control that is Automatically Taken Over from Failing OSA-ExpressIP Address Control that is Automatically Taken Over from Failing OSA-Express

In Figure 1, under normal conditions, OSA-Express CHPID F4 controls the network connection to IP addresses 10.10.1.1 and 10.10.1.2, while CHPID FC controls the connection to 10.10.1.3 and 10.10.1.4. If CHPID F4 fails, OSA-Express CHPID FC automatically takes control of IP addresses 10.10.1.1 and 10.10.1.2, while maintaining control of 10.10.1.3 and 10.10.1.4. These changes are made dynamically in the CHPID's OSA Address Table and communicated to the clients, alerting them to the new path.

The ARP cache for a targeted TCP/IP stack can be queried and purged from the z/OS console.

The qetharp utility is available on Linux to query and purge the ARP cache for a targeted TCP/IP stack.

Go to the previous page Go to the next page




Copyright IBM Corporation 1990, 2014