Document in Elasticsearch

What is an Elasticsearch document?

While an SQL database has rows of data stored in tables, Elasticsearch stores data as multiple documents inside an index. This is where the analogy must end however, since the way that Elasticsearch treats documents and indices differs significantly from a relational database.

For example, documents could be:

• Products in an e-commerce index
• Log lines in a data logging application
• Invoice lines in an invoicing system

Document fields

Each document is essentially a JSON structure, which is ultimately considered to be a series of key:value pairs. These pairs are then indexed in a way that is determined by the document mapping. The mapping defines the field data type as text, keyword, float, time, geo point or various other data types.

Elasticsearch documents are described as schema-less because Elasticsearch does not require us to pre-define the index field structure, nor does it require all documents in an index to have the same structure. However, once a field is mapped to a given data type, then all documents in the index must maintain that same mapping type.

Each field can also be mapped in more than one way in the index. This can be useful because we may want a keyword structure for aggregations, and at the same time be able to keep an analysed data structure which enables us to carry out full text searches for individual words in the field.

For a full discussion on mapping please see here.

Document source

An Elasticsearch document _source consists of the original JSON source data before it is indexed. This data is retrieved when fetched by a search query.

Document metadata

Each document is also associated with metadata, the most important items being:

_index – The index where the document is stored

_id – The unique ID which identifies the document in the index

Documents and index architecture

Note that different applications could consider  a “document” to be a different thing.  For example, in an invoicing system, we could have an architecture which stores invoices as documents (1 document per invoice),  or we could have an index structure which stores multiple documents as “invoice lines” for each invoice. The choice would depend on how we want to store, map and query the data.

Examples:

Creating a document in the user’s index:

POST  /users/_doc 
{
    "name" : "Petey",
    "lastname" : "Cruiser",
    "email" : "petey@gmail.com"
}

In the above request, we haven’t mentioned an ID for the document so the index operation generates a unique ID for the document. Here _doc is the type of document.

POST  /users/_doc/1
{
    "name" : "Petey",
    "lastname" : "Cruiser",
    "email" : "petey@gmail.com"
}

In the above query, the document will be created with ID 1.

You can use the below ‘GET’ query to get a document from the index using ID:

GET  /users/_doc/1

Below is the result, which contains the document (in _source field) as metadata:

{
    "_index": "users",
    "_type": "_doc",
    "_id": "1",
    "_version": 1,    "_seq_no": 1,    "_primary_term": 1,
    "found": true,
    "_source": {
        "name": "Petey",
        "lastname": "Cruiser",
        "email": "petey@gmail.com"
    }
}

Notes

Starting version 7.0 types are deprecated, so for backward compatibility on version 7.x all docs are under type ‘_doc’, starting 8.x type will be completely removed from ES APIs.

Overview

In Elasticsearch, routing refers to document routing. When you index a document, Elasticsearch will determine which shard will be used to index the document to. 

The shard is selected based on the following formula:

shard = hash(_routing) % number_of_primary_shards

Where the default value of _routing is _id.
It is important to know which shard the document is routed to, because Elasticsearch will need to determine where to find that document later on for document retrieval requests. 

Examples

In twitter index with 2 primary shards, the document with _id equal to “440” gets routed to the shard number:   

shard = hash( 440 ) % 2
PUT twitter/_doc/440
{
...
}

Notes and good things to know

• In order to improve search performance speed you can create custom routing. For example, you can enable custom routing that will ensure only a single shard is queried (the shard that contains your data).
• To create custom routing in Elasticsearch, you will need to configure and define that not all routing will be completed by default settings. ( v <= 5.0)

PUT my_index/customer/_mapping
{
   "order":{
      "_routing":{
         "required":true
      }
   }
}

•  This will ensure that every document in the “customer” type must specify a custom routing.  For elasticsearch 6 or above you will need to update the same mapping as:

PUT my_index/_mapping
{
   "order":{
      "_routing":{
         "required":true
      }
   }
}