The Master/Slave pattern is a very common design pattern in software engineering. Interestingly enough, there are two different patterns called Master/Slave, one primarily used in distributed computing, the other mostly found in the field of databases. There are some other uses of the term and discussions on the subject which are explained in the end.
Master/Slave: Shut Up and Follow
The most popular usa of 'Master/Slave' is in the field of databases. Here descibes a scenario where the slave database(s) replicate everything the master database does, often by reading and replicating all
UPDATE calls in the master database's binlog.
It follows that the master database can recieve both read and write requests while the slave database can only work with read requests. It also should be noted that since the slave replicates both
DELETE commands, a slave database is not sufficient as a backup.
Master/Slave: Divide and Conquer
In distributed computing, the Master/Slave pattern is also referred to as 'Divide and Conquer': the Master has large, repetitive task, so he splits it up and gives it to multiple Slave instances. They do the heavy work and the Master assembles the various returned results.
This is a simple pattern that has a variety of implementations, differing in how a task is split up, whether the master knows about his workers and/or the workers know about their master, if buffering is used to reduce idle times, how the tasks are transmitted etc.
Master/Slave in History
The term 'Master/Slave' has obvious imperialistic implications. Because of this, various groups interested in Political Correctness, amongst them the L.A. County and the Global Language Monitor have protested the use of these words. Both the Django and the Drupal project subsequently moved to use 'Primary/Replica' instead of 'Master/Slave'.
(Besides, 'Master/Slave' was also used in strange philosophical writings in the 19th century by both Hegel and Nietzsche. Just for the confused student of German literature who stubled upon this page.)