Not Analyzed & Multi-Fields

• Lucene and hence ElasticSearch break strings into terms using built-in or custom tokenizers

• Some strings don't make sense to tokenize e.g. uuid or guid often used as equivalent of a primary key and/or unique identifier

• not_analyzed: ElasticSearch mapping option to suppress tokenization:

  curl -XPUT 'localhost:9200/orders/orders/_mapping?pretty=true' \
   -H 'Content-Type: application/json' \
   -d '
  {
   "orders": {
      "properties": {
         "id": {
            "type": "text",
            "index": false
         }
      }
   }
  }'

• What you think difference will be searching or aggregating tokenized uuid/guid vs. non-tokenized uuid/guid property?

• What if I need both tokenized and non-tokenized option for the same field?

• Multi-Fields Mapping allows double indexing the same data:

  curl -XPUT 'localhost:9200/ordering/orders/_mapping?pretty=true' -d '
  {  
   "orders":{  
      "properties": {  
         "streetName": {
            "type":"text",
            "fields": {  
               "notparsed": {  
                  "type":"keyword",
                  "index":"not_analyzed"
               }
            }
         }
      }
   }
  }'