concept
The current backup software is more popular. The key technology used is IP address drift. In fact, the IP address drift technology uses a shared "virtual" IP address. When the local node switch and route switch are performed, the sender and receiver of the other end The end does not feel the change of the local node address, which ensures the continuity of communication and plays the role of breakpoint protection.
for example:
Assuming the simplest master-slave mode and single network card, the dual-system system will occupy 3 network addresses. The address of host one is 200.10.20.1, and the address of host two is 200.10.20.2. At the same time, there should also be a "virtual" IP address, which is a working IP address for "drifting", that is, the working address is 200.10.20.3 . When assigning IP addresses, distinguish between the working IP address and the host's own address. Under normal circumstances, the working host's own address is replaced by the working address (if host one is working, the address of host one is changed from 200.10.10.1 to 200.10.10.3). In the case of host switching, the address of host one will be restored to 200.10.10.1, and the address of host two will change from 200.10.10.2 to 200.10.10.3, which is what we call address drift.
Adopting address drifting technology and using virtual IP addresses to bind the main server can effectively solve the problem of information interaction between multiple server clusters when multicast cannot be used. This method has a certain degree of representativeness. In many industries, the dual-computer cluster method can be widely used in industrial control and backup of important events. The two central servers form a high-reliability dual-system hot backup system through the network, which works synchronously and backs up each other. When the host fails, it automatically switches to the backup machine, and the system runs uninterrupted. The address drift technology is when one of the hosts fails, the other host takes over the network IP address of the failed host, so that the network can still operate normally.