Data is replicated after it has been commited. Data loss can exist
if the master server crushes before the changes are commited by the remote server.
System has to ensure the data written by the transaction will be present
to at least two servers when the transaction commits. Confirmation
from the remote server is needed and this creates an overhead.
Single Master Replication
The master server replicates the data to the slave server.
The writes go to the Master server and these changes are disturbuted
to the slaves servers.
In this server writes are allowed to all the servers in the cluster.
This means alot of writes can go to many nodes at the same time.
This replication distributes the changes at a logical level. It
does not concern it self with physical layout of the data structure
of the database it self.
Is the type of replication where data are moved as is. The replication
of data is done at a binary level.