Multihoming enables an interface to assume multiple IP addresses on the same network. Multihoming can be used for all IP networks bound to a router. This is irrespective of whether the networks are bound to the same interface or to different interfaces. The most common use of multiple addresses on the same network is to enable a Web server to operate as though it were several Web servers. One application is to use each secondary IP address to point to a different Web page on the same Web server, depending on the Domain Name System (DNS) name that is used to reach the server.
Multihoming is also commonly used with network address translation (NAT), the proxy server, and the virtual private network (VPN). In all cases, the secondary IP address can be configured on the same interface that has the primary IP address. Or the secondary address can be configured on a different interface. When there are multiple interfaces, the secondary address is associated with the interface that is bound to the network that uses the same address. If the secondary address is not valid on any of the networks bound to existing interfaces, the address is rejected and an error message is produced.
With this release, the multihoming feature is extended to help configure the stack for load balancing and fault tolerance at NIC/Link level. The TCP/IP stack also provides mechanisms for the administrator to group those NICs which are similar in characteristics to facilitate load balancing and fault tolerance across them.
For NetWare 6, TCP/IP has two levels of enabling load balancing and fault tolerance, one at the system level and another at the local group level. To benefit from the feature make sure that you have fully enabled it, at both the levels.
Five types of multihoming configurations are provided:
Single NIC Multiple IP Addresses (Different IP Addresses)---In this type of multihoming the client to server traffic can be distributed across the routers. The required level of load balancing can be achieved through static routers and dynamic DNS. Any dynamic algorithm would take care of the fault tolerance aspect. In the case of a static route configuration the Multiple Default Gateway with Dead Gateway Detection support would also help achieve the same results.
Multiple NIC Multiple IP Address (Different IP Network)---This is a normal configuration of a router. All the configuration and the advantages gained by the previous type of multihoming are applicable here also. In addition, load balancing server-to-client traffic across the NICs and routers is also possible. This can be achieved with the help of multipath routing.
Multiple NIC Multiple IP Address (Same IP Network)---This type of multihoming allows for both server-to-client and client-to-server load balancing and fault tolerance at the Link level. This type of multihoming is especially helpful in those cases where Route level load balancing and fault tolerance are not required and become an overhead. Once this type of multihoming is supported on the server side, the outgoing traffic load is NIC based on various parameters such as the destination IP address and interference lead. During NIC failure, the lead is automatically distributed among the remaining NICs to achieve fault tolerance.
Multiple NIC Single IP Address---Here the clients use the same IP address to communicate with the server. This is achieved through using the round robin method of distributing the NIC addresses for the ARP requests sent by the clients. This solution avoids the extra configuration and transparently achieves the load balancing for incoming packets. During a NIC failure, the server sends a message to the clients to forcefully use them as the other interface's MAC address.
Secondary IP Address (Multiple Logical Hosts)---With this type of multihoming it is possible to create multiple logical hosts belonging to the same network. In a multihoming setup where multiple NICs are grouped to support a single network, the secondary IP address with this type of configuration supports an option to select one of the NICs in the group. By using the non-ARPable option, these addresses can be used as virtual IP addresses for load balancing solutions. Through this option, the same IP address can be configured on all servers and the load balancer can distribute the client load across these servers.
For more information on this, see Configuring Load Balancing and Configuring Fault Tolerance.