Redundancy – using more elements than minimally required. For example, if a server has only one common element running, the entire system will fail if it fails. If the server is provided with additional redundant resources, they will replace the damaged one and prevent the server from stopping. Providing redundancy is possible in several ways.
This includes additional drives and power supplies. During a failure of one of the servers, for example, the data is distributed to multiple drives at the same time. In addition, it is necessary to set up backup of information in several places.
Duplication of data on the server is carried out in several ways:
- mirroring. With special mirroring modes, channels are duplicated. If one fails, the other one is activated;
- disks. Here the principle of operation is the same as that of mirroring. The information is simply written to two disks: the original and its copy;
- power supply. Servers are equipped with at least two power supplies. If one is damaged, the server continues to function at the expense of the other.
Thus, redundancy is achieved by additional nodes. They are mounted according to a certain scheme using N, where N is the load element, which should be sufficient for continuous operation of the infrastructure.
N+1. A single redundant element is used here. At the moment of a system failure, this component will take over the entire load. Variations of N+2, N+3, etc. are possible.
2N – two parallel, redundant systems, working for each individual N. As in the first case, 2N, 3N are additional redundant elements.
Thus, redundancy is provided by the additional N. If 1-2 components fail, the load will be transferred to the redundant ones, they can be easily replaced without stopping the work. It should be understood that each additional element, although it provides additional redundancy, but may be unclaimed.