_tier field

_tier field

When performing queries across multiple indexes, it is sometimes desirable to target indexes held on nodes of a given data tier (data_hot, data_warm, data_cold or data_frozen). The _tier field allows matching on the tier_preference setting of the index a document was indexed into. The preferred value is accessible in certain queries :

  1. resp = client.index(
  2.     index="index_1",
  3.     id="1",
  4.     document={
  5.         "text": "Document in index 1"
  6.     },
  7. )
  8. print(resp)
  10. resp1 = client.index(
  11.     index="index_2",
  12.     id="2",
  13.     refresh=True,
  14.     document={
  15.         "text": "Document in index 2"
  16.     },
  17. )
  18. print(resp1)
  20. resp2 = client.search(
  21.     index="index_1,index_2",
  22.     query={
  23.         "terms": {
  24.             "_tier": [
  25.                 "data_hot",
  26.                 "data_warm"
  27.             ]
  28.         }
  29.     },
  30. )
  31. print(resp2)
  1. response = client.index(
  2.   index: 'index_1',
  3.   id: 1,
  4.   body: {
  5.     text: 'Document in index 1'
  6.   }
  7. )
  8. puts response
  10. response = client.index(
  11.   index: 'index_2',
  12.   id: 2,
  13.   refresh: true,
  14.   body: {
  15.     text: 'Document in index 2'
  16.   }
  17. )
  18. puts response
  20. response = client.search(
  21.   index: 'index_1,index_2',
  22.   body: {
  23.     query: {
  24.       terms: {
  25.         _tier: [
  26.           'data_hot',
  27.           'data_warm'
  28.         ]
  29.       }
  30.     }
  31.   }
  32. )
  33. puts response
  1. const response = await client.index({
  2.   index: "index_1",
  3.   id: 1,
  4.   document: {
  5.     text: "Document in index 1",
  6.   },
  7. });
  8. console.log(response);
  10. const response1 = await client.index({
  11.   index: "index_2",
  12.   id: 2,
  13.   refresh: "true",
  14.   document: {
  15.     text: "Document in index 2",
  16.   },
  17. });
  18. console.log(response1);
  20. const response2 = await client.search({
  21.   index: "index_1,index_2",
  22.   query: {
  23.     terms: {
  24.       _tier: ["data_hot", "data_warm"],
  25.     },
  26.   },
  27. });
  28. console.log(response2);
  1. PUT index_1/_doc/1
  2. {
  3.   "text": "Document in index 1"
  4. }
  6. PUT index_2/_doc/2?refresh=true
  7. {
  8.   "text": "Document in index 2"
  9. }
  11. GET index_1,index_2/_search
  12. {
  13.   "query": {
  14.     "terms": {
  15.       "_tier": ["data_hot", "data_warm"] 
  16.     }
  17.   }
  18. }

Querying on the _tier field

Typically a query will use a terms query to list the tiers of interest but you can use the _tier field in any query that is rewritten to a term query, such as the match, query_string, term, terms, or simple_query_string query, as well as prefix and wildcard queries. However, it does not support regexp and fuzzy queries.

The tier_preference setting of the index is a comma-delimited list of tier names in order of preference i.e. the preferred tier for hosting an index is listed first followed by potentially many fall-back options. Query matching only considers the first preference (the first value of a list).