That you need to take care of regular copies of important data in general is already obvious.
In a significant proportion of cases, the solution which in the event of failure
Efficient repair or replacement of a damaged server and restoration of data from a copy is a sufficient solution.

 

We risk an outage of several hours possibly re-entering the data we entered from the creation of a regular copy to the failure.

 

Treat sometimes the standard approach is not enough.
In some situations with a certain scale of services provided, for example, online stores or services used by a large number of users begin to have completely different requirements.

 

- high availability systems - fault tolerance
In such cases, any failure that causes services to stop is a real financial loss because, for example, customers cannot shop or use services all the more the loss of information about transactions made is unopenable. Not to mention image losses.

 

Then we build the infrastructure so that the failure of any single component cannot stop the service.

 

We prepare scenarios of action no longer only what to do: when a disk drive or power supply breaks down, but also what to do when the whole server breaks down, through problems with network failures, power supply in the server room to such extreme scenarios as a fire of the entire server room in which we have servers ... it has already happened ... it has already happened.

 

Scalability
A separate series of issues at a certain scale of users are the capabilities of individual devices which always run out at some point.

 

In such situations, we configure an infrastructure capable of flexibly distributing the load so as to support the number of users that often far exceeds the capacity of a single server.

 

Also, it is a matter of time with a certain scale of services to face different types of attacks on the infrastructure for which you also need to be prepared.

 

Infrastructure monitoring then becomes very important, allowing us to react in real time to incidents and problems before users can even feel them.