When an IPv4 DYNAMICXCF HiperSockets device and link are created and successfully activated, a subnet route is created across the HiperSockets link. The subnet is created by using the DYNAMICXCF IP address and mask. This allows any LPAR within the same CPC to be reached, even ones that are not within the sysplex. For example, an LPAR that is running z/Linux or z/VM® that does not support joining the sysplex can still be reached. The z/Linux or z/VM LPAR must define at least one IP address for the HiperSockets endpoint that is within the subnet defined by the DYNAMICXCF IP address and mask.
Similarly, when an IPv6 DYNAMICXCF HiperSockets interface is created and successfully activated, a prefix route is created across the HiperSockets interface (if prefix_route_len is specified on DYNAMICXCF). This allows any LPAR within the same CPC to be reached, even ones that are not within the sysplex. For example, an LPAR that is running z/Linux or z/VM that does not support joining the sysplex can still be reached. The z/Linux or z/VM LPAR must define at least one IP address for the HiperSockets endpoint that uses the same prefix as the DYNAMICXCF IP address.
Therefore, TCP/IP can communicate with other LPARs within the CPC over the HiperSockets connectivity created by DYNAMICXCF even when the TCP/IP in the other LPAR is not part (joins or supports) of the sysplex. You can also elect to manually configure a HiperSockets device for non-sysplex communications.
It is also possible with multiple stacks in the same LPAR to end up with both XCF and HiperSockets links or interfaces. This occurs when the availability of the (preferable) HiperSockets link or interface changes as each TCP stack (within the same LPAR) is started. For example, stack A is started with HiperSockets available and later stack B is started with HiperSockets unavailable. This type of configuration should be avoided.