Deploying a Cluster on Google Cloud Platform

Deploy Hazelcast clusters on Google Cloud Platform and allow them to discover each other automatically.

Before you Begin

Before deploying Hazelcast on GCP, your GCP instances must have access to the Cloud API (at minimum read-only access scope to the Compute Engine API).

Discovering Members Automatically

To make it easier to set up clusters on GCP, Hazelcast allows members to discover each other automatically, using discovery strategies.

When a member starts on GCP, members start by fetching a list of all instances filtered by projects, zones, and label. Then, each instance is checked one-by-one with its IP and each of the ports defined in the hz-port property. When a member is discovered under IP:PORT, then it joins the cluster.

If you want to create multiple Hazelcast clusters in one project/zone, then you need to manually label the instances.

Member Discovery SPI

You can configure Hazelcast in one of the following manners.

XML
YAML
Java

<hazelcast>
  <network>
    <join>
      <multicast enabled="false"/>
      <gcp enabled="true">
        <label>application=hazelcast</label>
        <hz-port>5701-5708</hz-port>
      </gcp>
    </join>
  </network>
</hazelcast>

hazelcast:
  network:
    join:
      multicast:
        enabled: false
      gcp:
        enabled: true
        label: application=hazelcast
        hz-port: 5701-5708

config.getNetworkConfig().getJoin().getMulticastConfig().setEnabled(false);
config.getNetworkConfig().getJoin().getGcpConfig().setEnabled(true)
  .setProperty("label", "application=hazelcast")
  .setProperty("hz-port", "5701-5708");

The following properties can be configured:

private-key-path: a filesystem path to the private key for GCP service account in the JSON format; if not set, the access token is fetched from the GCP VM instance
projects: a list of projects where the plugin looks for instances; if not set, the current project is used
- region: a region where the plugin looks for instances; if not set, the zones property is used; if it and zones property not set, all zones of the current region are used
zones: a list of zones where the plugin looks for instances; if not set, all zones of the current region are used
label: a filter to look only for instances labeled as specified; property format: key=value
hz-port: a range of ports where the plugin looks for Hazelcast members; if not set, the default value 5701-5708 is used

Note that:

Your GCP Service Account must have permissions to query for all the projects/zones specified in the configuration
If you don’t specify any of the properties, then the plugin forms a cluster from all Hazelcast members running in the current project, in the current region
If you use the plugin in the Hazelcast Client running outside the GCP network, then the following parameters are mandatory: private-key-path, projects, and zones or region

Preventing Data Loss

By default, Hazelcast distributes partition replicas (backups) randomly and equally among cluster members. However, this is not safe in terms of high availability when a partition and its replicas are stored on the same rack, using the same network, or power source. To deal with that, Hazelcast offers logical partition grouping, so that a partition itself and its backups would not be stored within the same group. This way Hazelcast guarantees that a possible failure affecting more than one member at a time will not cause data loss. For more details about partition groups, see Partition Group Configuration.

Zone Aware

When using ZONE_AWARE configuration, backups are created in the other availability zone.

XML
YAML
Java

<partition-group enabled="true" group-type="ZONE_AWARE" />

hazelcast:
  partition-group:
    enabled: true
    group-type: ZONE-AWARE

config.getPartitionGroupConfig()
    .setEnabled(true)
    .setGroupType(MemberGroupType.ZONE_AWARE);

When using the ZONE_AWARE partition grouping, a cluster spanning multiple availability zones should have an equal number of members in each AZ. Otherwise, it will result in uneven partition distribution among the members.

Discovering Members from Hazelcast Clients

If you run Hazelcast clients inside GCP, then the configuration is exactly the same as for the members.

If you run Hazelcast clients outside GCP, then you always need to specify the following parameters in your client configuration:

private-key-path - path to the private key for GCP service account
projects - a list of projects where the plugin looks for instances
region: a region where the plugin looks for instances; if not set, the zones property is used; if it and zones property not set, all zones of the current region are used
zones: a list of zones where the plugin looks for instances; if not set, all zones of the current region are used
use-public-ip - must be set to true

The following snippets are example declarative and programmatic configurations.

XML
YAML
Java

<hazelcast-client>
  <network>
    <gcp enabled="true">
      <private-key-path>/home/name/service/account/key.json</private-key-path>
      <projects>project-1,project-2</projects>
      <region>us-east1</region>
      <label>application=hazelcast</label>
      <hz-port>5701-5708</hz-port>
      <use-public-ip>true</use-public-ip>
    </gcp>
  </network>
</hazelcast-client>

hazelcast-client:
  network:
    gcp:
      enabled: true
      private-key-path: /home/name/service/account/key.json
      projects: project-1,project-2
      region: us-east1
      label: application=hazelcast
      hz-port: 5701-5708
      use-public-ip: true

clientConfig.getGcpConfig().setEnabled(true)
      .setProperty("private-key-path", "/home/name/service/account/key.json")
      .setProperty("projects", "project-1,project-2")
      .setProperty("region", "us-east1")
      .setProperty("label", "application=hazelcast")
      .setProperty("hz-port", "5701-5708")
      .setProperty("use-public-ip", "true");