Install on CentOS
Installation on CentOS 7 or 8.
- Tune your operating system for optimal performance.
EPAS version 12.7 or 13.3 installed, see https://www.enterprisedb.com/downloads/edb-postgres-advanced-server.Note: To find and install a specific minor version of EPAS do the following:
sudo yum --show-duplicates list <package name> sudo yum install <package name>-<version info>
For CentOS 7 you have to additionally install the following repository:
sudo yum install -y epel-release centos-release-scl
Install the Swarm64 DA repository:
curl -s https://packagecloud.io/install/repositories/swarm64/swarm64da/script.rpm.sh | sudo bash
Install the corresponding Swarm64 DA package for your EPAS installation.
- EPAS 12
sudo yum install swarm64da-epas-12
- EPAS 13
sudo yum install swarm64da-epas-13
shared_preload_librariesin your postgresql.conf configuration file.For example:
For other recommended settings see also Recommended EPAS settings.
shared_preload_libraries = 'swarm64da'
- Restart EPAS.
Switch to a user with access to the database:
sudo su - enterprisedb
Connect to the database:
edb-psql -d postgres
Load the Swarm64 DA extension:
CREATE EXTENSION swarm64da;
NOTICEis printed when executing the first SQL command through a connection to a database for which the
swarm64daextension has not been created yet. This behavior is controlled by the following configuration parameter:
- swarm64da.enable_notice_not_created_extension (boolean)
Enables or disables the reminder to create the extension. The default is
Verify that the Swarm64 DA extension is loaded correctly and look up its
SELECT * FROM swarm64da.get_version();
Install a valid license.
See also Install a license for Swarm64 DA.
- Check and follow the Swarm64 DA settings advisor and documented recommendations to tune your database system for optimal performance.
Create columnstore indexes.
Columnstore indexes are necessary to maximize the performance benefits. Create an index for each large table covering the most common accessed columns. See Columnstore index and Using a columnstore index on a small TPC-H data set.