Creating a Cluster on AWSPDF version

Placement group support

You can configure a Cloudera Data Hub cluster on AWS for placement group support. Placement groups support placing VM instances across different physical hardware within a single availability zone. Placement groups can help ensure cluster availability in the event of a physical hardware failure within an availability zone.

In a Cloudera Data Hub cluster, each host group can be associated with a placement group, of which there are three supported types: partition, spread, or cluster. Cloudera recommends using the partition type as the default for all host groups. For a partition placement group, the partition count will always be 2 and is not configurable.

Configuring placement group support requires adding custom properties to a cluster definition at the host group level. You can define the strategy as “NONE,” “PARTITION,” “SPREAD,” or “CLUSTER.”

For example:

{
  "environmentName": "sample-env",
  "instanceGroups": [
    {
      "nodeCount": 1,
      "name": "master",
      "type": "GATEWAY",
      "recoveryMode": "MANUAL",
      "template": {
        "aws": {
          "encryption": {
            "type": "NONE",
            "key": null
          },
          "placementGroup": {
            "strategy": "PARTITION"
          }
        },
        "instanceType": "m5.2xlarge",
        "rootVolume": {
          "size": 50
        },
        "attachedVolumes": [
          {
            "size": 100,
            "count": 1,
            "type": "standard"
          }
        ],
        "cloudPlatform": "AWS"
      },

Placement groups have a number of rules and limitations. Importantly, it is possible to run out of placement groups within an availability zone. Refer to the AWS documentation for detailed rules and limitations.

We want your opinion

How can we improve this page?

What kind of feedback do you have?