In the previous chapter, we've seen how to build several slaves from a master node. When you have a few nodes, it's not really a problem for the master node to handle them.
However, when you multiply the number of slave nodes, you'll encounter more and more I/O problems with your master node. It will be hard for it to perform write operations and give them to the slave nodes from the binary logs. So basically, you shouldn't have these kinds of architectures.
Refer to the following diagram:
Monitor the master node properly to ensure that it is not overloaded. Of course, the number of maximum slaves depends on the load of the master node.
To resolve this kind of problem, you can perform slave cascading. The following diagram shows a simple solution:
This means that you can have a master replicated on a slave and then replicated on another slave (as shown in the following diagram). This really helps in scaling with high load traffic.
