Consider a two-tier architecture in which there is a User Interface (UI) instance that connects to a database instance. Now suppose that the system gets a large burst of traffic. Using the scale out pattern discussed in Chapter 2, Basic Patterns, the operations team can manually or automatically add more UI instances to the load balancer to mitigate the flux in traffic. This server redundancy pattern is the multi-server pattern. This pattern includes a healthcheck on the instance so that, if one of the instances that the load is being distributed across is misbehaving, it will not continue to route into it.
Note
One might note that this pattern is similar to the scale out pattern except that it is not dynamic in nature. In this pattern, we are discussing the details of adding horizontal scalability in a much simpler example.
A few things should be noted before we continue, however. It has been found, in my personal usage as well as documented cases, that the servers should...