Amazon Elasticsearch Service
Developer Guide (API Version 2015-01-01)

Indexing Data in Amazon Elasticsearch Service

Because Elasticsearch uses a REST API, numerous methods exist for indexing documents. You can use standard clients like curl or any programming language that can send HTTP requests. To further simplify the process of interacting with it, Elasticsearch has clients for many programming languages. Advanced users can skip directly to Programmatic Indexing.

For situations in which new data arrives incrementally (for example, customer orders from a small business), you might use the _index API to index documents as they arrive. For situations in which the flow of data is less frequent (for example, weekly updates to a marketing website), you might prefer to generate a file and send it to the _bulk API. For large numbers of documents, lumping requests together and using the _bulk API offers superior performance. If your documents are enormous, however, you might need to index them individually using the _index API.

For information about integrating data from other AWS services, see Loading Streaming Data into Amazon Elasticsearch Service.

Introduction to Indexing

Before you can search data, you must index it. Indexing is the method by which search engines organize data for fast retrieval. The resulting structure is called, fittingly, an index.

In Elasticsearch, the basic unit of data is a JSON document. Within an index, Elasticsearch organizes documents into types (arbitrary data categories that you define) and identifies them using a unique ID.

A request to the _index API looks like the following:

PUT elasticsearch_domain/index/type/id { "A JSON": "document" }

A request to the _bulk API looks a little different, because you specify the index, type, and ID in the bulk data:

POST elasticsearch_domain/_bulk { "index": { "_index" : "index", "_type" : "type", "_id" : "id" } } { "A JSON": "document" }

Bulk data must conform to a specific format, which requires a newline character (\n) at the end of every line, including the last line. This is the basic format:

action_and_metadata\n optional_document\n action_and_metadata\n optional_document\n ...

For a short sample file, see Step 2: Upload Data to an Amazon ES Domain for Indexing.

Elasticsearch features automatic index creation when you add a document to an index that doesn't already exist. It also features automatic ID generation if you don't specify an ID in the request. This simple example automatically creates the movies index, establishes the document type of movie, indexes the document, and assigns it a unique ID:

curl -XPOST elasticsearch_domain/movies/movie -d '{"title": "Spirited Away"}' -H 'Content-Type: application/json'


To use automatic ID generation, you must use the POST method instead of PUT.

To verify that the document exists, you can perform the following search:

curl -XGET elasticsearch_domain/movies/_search?pretty

The response should contain the following:

"hits" : { "total" : 1,   "max_score" : 1.0,   "hits" : [   {     "_index" : "movies",       "_type" : "movie",       "_id" : "AV4WaTnYxBoJaZkSFeX9",       "_score" : 1.0,       "_source" : {         "title" : "Spirited Away"       }     }   ] }

Automatic ID generation has a clear downside: because the indexing code didn't specify a document ID, you can't easily update the document at a later time. To specify an ID of 7, use the following request:

curl -XPUT elasticsearch_domain/movies/movie/7 -d '{"title": "Spirited Away"}' -H 'Content-Type: application/json'

Indices default to five primary shards and one replica. If you want to specify non-default settings, create the index before adding documents:

curl -XPUT elasticsearch_domain/movies -d '{"settings": {"number_of_shards": 6, "number_of_replicas": 2}}' -H 'Content-Type: application/json'


Requests using curl are unauthenticated and rely on an IP-based access policy. For examples of signed requests, see Programmatic Indexing.

Elasticsearch indices have the following naming restrictions:

  • All letters must be lowercase.

  • Index names cannot begin with _ or -.

  • Index names cannot contain spaces, commas, ", *, +, /, \, |, ?, >, or <.

On this page: