PGD-Always-ON v23

EDB Postgres Distributed 5 in an Always-ON configuration, suitable for use in test and production.

This architecture is valid for use with EDB Postgres Distributed version 5 only and requires a subscription to EDB Repos 2.0.

Cluster configuration

Overview of configuration options

This example shows an invocation of tpaexec configure for this architecture:

tpaexec configure ~/clusters/pgd-ao \
         --architecture PGD-Always-ON \
         --edb-postgres-extended 15 \
         --platform aws --instance-type t3.micro \
         --distribution Debian \
         --pgd-proxy-routing global \
         --location-names dc1 dc2 dc3 \
         --witness-only-location dc3 \
         --data-nodes-per-location 2

You can list all available options using the help option.

tpaexec configure --architecture PGD-Always-ON --help

Mandatory options

--architecture (-a)Must be set to PGD-Always-ON
Postgres flavour and version (e.g. --postgresql 15)A valid flavour and version specifier.
--pgd-proxy-routingMust be either global or local.

Additional options

OptionsDescriptionBehavior if omitted
--platformOne of aws, docker, bare.Defaults to aws.
--location-namesA space-separated list of location names. The number of locations is equal to the number of names supplied.TPA will configure a single location with three data nodes.
--witness-only-locationA location name, must be a member of location-names.No witness-only location is added.
--data-nodes-per-locationThe number of data nodes in each location, must be at least 2.Defaults to 3.
--add-proxy-nodes-per-locationThe number of proxy nodes in each location.PGD-proxy will be installed on each data node.
--enable-camoSets two data nodes in each location as CAMO partners.CAMO will not be enabled.
--bdr-databaseThe name of the database to be used for replication.Defaults to bdrdb.
--enable-pgd-probesEnable http(s) API endpoints like health/is-ready to probe pgd-proxy's health.Disabled by default.

More detail about PGD-Always-ON configuration

A PGD-Always-ON cluster comprises a number of locations, preferably odd. Each location has the same number of data nodes, also preferably odd. If you don't specify any --location-names, the default is to use a single location with three data nodes.

You can specify location names for the cluster with --location-names dc1 dc2 …. A location represents an independent data center that provides a level of redundancy, in whatever way this definition makes sense to your use case. Examples include AWS regions, your own data centers, or any other designation to identify where your servers are hosted.

for AWS users

If you're using TPA to provision an AWS cluster, the locations will be mapped to separate availability zones in the --region you specify. You can specify multiple --regions, but TPA doesn't currently set up VPC peering to allow instances in different regions to communicate with each other. For a multi-region cluster, you must set up VPC peering yourself.

Use --data-nodes-per-location N to specify the number of data nodes in each location. The minimum number is 2. The default is 3.

If you specify an even number of data nodes per location, TPA adds an extra witness node to each location. The witness preserves the ability to reliably establish consensus without the hardware requirements and extra cost of another data node.

A cluster with only two locations would entirely lose the ability to establish global consensus if one of the locations were to fail. You can retain the ability to establish consensus despite a single-location failure. To do so, we recommend adding a third witness-only location that contains only a witness node. Use --witness-only-location loc to designate one of your locations as a witness.

By default, every data node in every location also runs PGD Proxy for connection routing. To create separate PGD Proxy instances instead, use --add-proxy-nodes-per-location 3 (or however many proxies you want to add).

Depending on your use case, you must specify --pgd-proxy-routing local or global to configure how PGD Proxy routes connections to a write leader. Local routing makes every PGD Proxy route to a write leader within its own location (suitable for geosharding applications). Global routing makes every proxy route to a single write leader, elected among all available data nodes across all locations.

You can optionally specify:

  • --bdr-database dbname to set the name of the database with BDR enabled (default: bdrdb)

  • --enable-camo to set two data nodes in each region as CAMO partners

  • --enable-pgd-probes [{http, https}] to enable http(s) api endpoints that allow you to easily probe proxy's health

You can also specify any of the options described by tpaexec help configure-options.