<!-- $PostgreSQL: pgsql/doc/src/sgml/high-availability.sgml,v 1.10 2006/11/22 04:00:19 momjian Exp $ -->

<!-- $PostgreSQL: pgsql/doc/src/sgml/high-availability.sgml,v 1.11 2006/11/22 04:01:40 momjian Exp $ -->

<chapter id="high-availability">

<title>High Availability and Load Balancing</title>

</varlistentry>

<varlistentry>

<term>Master/Slave Replication</term>

<term>Master-Slave Replication</term>

<listitem>

<para>

A master/slave replication setup sends all data modification

A master-slave replication setup sends all data modification

queries to the master server. The master server asynchronously

sends data changes to the slave server. The slave can answer

read-only queries while the master server is running. The

</varlistentry>

<varlistentry>

<term>Multi-Master Replication</term>

<term>SynchonousMulti-Master Replication</term>

<listitem>

<para>

In multi-master replication, each server can accept write

requests, and modified data is transmitted from the original

server to every other server before each transaction commits.

Heavy write activity can cause excessive locking, leading to

poor performance. In fact, write performance is often worse

than that of a single server. Read requests can be sent to

any server. Some implementations use cluster-wide shared memory

or shared disk to reduce the communication overhead. Clustering

is best for mostly read workloads, though its big advantage is

that any server can accept write requests &mdash; there is no

need to partition workloads between master and slave servers,

and because the data changes are sent from one server to another,

there is no problem with non-deterministic functions like

<function>random()</>.

Insynchonousmulti-master replication, each server can accept

writerequests, and modified data is transmitted from the

originalserver to every other server before each transaction

commits.Heavy write activity can cause excessive locking,

leading topoor performance. In fact, write performance is

often worsethan that of a single server. Read requests can

be sent toany server. Some implementations use cluster-wide

shared memoryor shared disk to reduce the communication

overhead. Clusteringis best for mostly read workloads, though

its big advantage isthat any server can accept write requests

&mdash; there is noneed to partition workloads between master

andslave servers, andbecause the data changes are sent from

one server to another,there is no problem with non-deterministic

functions like<function>random()</>.

</para>

<para>

</varlistentry>

<varlistentry>

<term>Multi-MasterWith Conflict Resolution</term>

<term>AsynchronousMulti-MasterReplication</term>

<listitem>

<para>