Decoupling Index Storage and Compute in an Elasticsearch Cluster Through Index Lifecycle Management

Overview

CSS supports decoupled storage and compute. That is, indexes can be frozen in OBS to reduce the storage cost of cold data. This document describes how to use index lifecycle management to automatically freeze indexes at a specific time to decouple storage and compute.

In this section, a lifecycle policy is configured to automatically freeze an index three days after it is created and dump data to OBS. The index will be deleted seven days after it is created.

Prerequisites

There are available CSS clusters.
Elasticsearch 7.6.2 clusters or later are used.

Decoupling Index Storage and Compute Through Index Lifecycle Management

Log in to the CSS management console.
In the navigation tree on the left, choose Clusters > Elasticsearch. The cluster list is displayed.
Click Access Kibana in the Operation column of a cluster.
In the navigation tree on the left of Kibana, choose Dev Tools. The command execution page is displayed.

Create a lifecycle policy named hot_warm_policy.

Policy description: Three days after an index is created, the API for freezing the index is automatically called to dump data to OBS. Seven days after an index is created, the index is deleted.

PUT _opendistro/_ism/policies/hot_warm_policy
{
  "policy": {
    "description": "hot warm delete workflow",
    "error_notification": null,
    "default_state": "hot",
    "states": [
      {
        "name": "hot",
        "actions": [],
        "transitions": [
          {
            "state_name": "warm",
            "conditions": {
              "min_index_age": "3d"
            }
          }
        ]
      },
      {
        "name": "warm",
        "actions": [
          {
            "freeze_low_cost": {}
          }
        ],
        "transitions": [
          {
            "state_name": "delete",
            "conditions": {
              "min_index_age": "7d"
            }
          }
        ]
      },
      {
        "name": "delete",
        "actions": [
          {
            "delete": {}
          }
        ],
        "transitions": []
      }
    ]
  }
}

Create the index template template_hot_warm.

Template description: All the new indexes starting with data are automatically associated with the lifecycle policy hot_warm_policy.

PUT _template/template_hot_warm
{
  "index_patterns": "data*",
  "settings": {
    "number_of_replicas": 5,
    "number_of_shards": 1,
    "opendistro.index_state_management.policy_id": "hot_warm_policy"
  },
  "mappings": {
    "properties": {
      "name": {
        "type": "text"
      }
    }
  }
}

**Table 1** Parameter description
Parameter	Description
number_of_shards	Number of index shards
number_of_replicas	Number of index shard replicas
opendistro.index_state_management.policy_id	Lifecycle policy name

Create the data-2022-06-06 index. The index automatically uses the template_hot_warm template and associates the index template with the lifecycle policy hot_warm_policy. In this way, the index is frozen three days after creation and is deleted seven days after creation.
```
POST data-2022-06-06/_bulk
{"index":{}}
{"name":"name1"}
{"index":{}}
{"name":"name2"}
{"index":{}}
{"name":"name3"}
{"index":{}}
{"name":"name4"}
{"index":{}}
{"name":"name5"}
{"index":{}}
{"name":"name6"}
```
Query data and check whether storage and compute is automatically decoupled.
- Three days after the index is created, check the frozen index.
```
GET _cat/freeze_indices?s=i&v
```
  The index generated three days ago is expected to be frozen.
```
health status index                  uuid                   pri rep docs.count docs.deleted store.size pri.store.size
green  open   data-2022-06-06  x8ab5NX6T3Ox_xoGUanogQ    1   1          6            0      7.6kb          3.8kb
```
- Seven days after the index is created, check the frozen index. The index is expected to be deleted.