Database Reference
In-Depth Information
CHAPTER 10
Replication and External Data
PostgreSQL has a number of options for sharing data with external servers or data
sources. The first option is the built-in replication options of PostgreSQL, which allows
you to create a copy of your server ready to run on another PostgreSQL server. The
second option is to use third-party add-ons, many of which are freely available and
time-tested. The third option, unveiled in version 9.1, is to use
foreign data wrappers
(FDW). FDWs gives you the flexibility to query from a wide array of external data
permit updating.
Replication Overview
The seemingly countless reasons for the need to replicate your databases all distill down
to two: availability and scalability. If your main server goes down, you want another to
immediately assume its role. For small databases, you could just make sure you have
another physical server ready and restore the database onto it. But for large databases
(in the terabytes), the restore itself could take many hours. To avoid downtime, you'll
need to replicate. The other main reason is scalability. You set up a database to breed
fancy
elephant shrews
for profit. After a few years of breeding, you now have thousands
of shrews. People all over the world come to your site to gawk and purchase. You're
overwhelmed by the traffic. Replication comes to your aid; you set up a read-only slave
server to replicate with your main server. You direct the countless gawkers to the slave,
and only let serious buyers onto the master server to finalize their purchases.
Replication Jargon
Before we get too carried away with replication, we had better introduce some common
lingo used in connection with it:
