defines structure of docs and how values are indexes
--> similar to schema in RDB

explicit mapping : defined by developer

dynamic mapping : auto generate field mapping

combining two are possible

1. Explicit Mapping

field mappings can be created when creating an index or afterwards

PUT /reviews
{
    "mappings": {
        "properties": {
            "rating": {"type": "float"},
            "content": {"type": "text"},
            "product_id": {"type": "integer"},
            "author": {
                "properties": {
                    "first_name": {"type": "text"},
                    "last_name": {"type": "text"},
                    "email": {"type": "keyword"}
                }
            }
        }
    }
}

2. Retrieving Mapping

for entire index

GET /reviews/_mapping

for specific field

GET /reviews/_mapping/field/content

// for field in object
GET /reviews/_mapping/field/author.email

3. Using dot notation in field names

PUT /reviews
{
    "mappings": {
        "properties": {
            "rating": {"type": "float"},
            "content": {"type": "text"},
            "product_id": {"type": "integer"},
            "author.first_name": {"type": "text"}
            "author.last_name": {"type": "text"}
            "author.email": {"type": "keyboard"}
        }
    }
}

dot notation can be used insted of nesting for better viewability

4. Adding Mappings to Existing Indexes

// add time stamp to reviews index
PUT /reviews/_mapping
{
    "properties": {
        "created_at": {"type": "date"}
    }
}

Date data type

specifying date

formatted string

milliseconds since the epoch (long)

seconds since the epoch (int)

custom date format

epoch = 1/1/1970

default behavior of date fields

Accepted formats

date without time

date with time

milliseconds since epoch (long)

UTC assumed

Date must be formatted according to ISO 8601 spec

how date fields are stored

stored internally as milliseconds since epoch

PUT /reviews/_doc/3
{
    "created_at": "2015-04-15T13:07:41Z"
}

how missing fields are handled

all fields are optional

unlike RDB where NULL allow should be explicitly set

integrity checks should be done at application level