Home

Awesome

elasticsearch-tableau-connector

Tableau Web Data Connector for Elasticsearch

Overview

This is an instance of a Tableau Web Data Connector for Elasticsearch. This will extract a set of data from Elasticsearch based on the cluster URL, index name, type and optional custom query and create a Tableau extract of the data. The connector will query the specified type's mapping in Elasticsearch to report the fields and data types that Tableau can expect to see.

This connector should be periodically refreshed, as the extract only includes data from the point in time that it was executed.

The connector works by retrieving 'pages' of data from Elasticsearch up to either the limit specified, or up to the total number of hits. The user can override the batch size to retrieve more records per page if desired.

Demo

Live Demo of Connector

Compatibility

The 2.3 release (in the release-2.3 branch and current development version in master) supports Tableau 10.4 or later.

The 2.0 release (in the release-2.0 branch) supports Tableau 10.0 or later.

The 1.0 release (in the release-1.0 branch) supports Tableau 9.1.6 or later, 9.2.4 or later, and 9.3.

Elasticsearch 5.+ is recommended

Known Issues and Limitations

Configuration

There is some configuration needed in Elasticsearch for the connector to work:

Enable CORS

http.cors.enabled: true

Additionally, in current versions of Elasticsearch (2.3+), it is required to define which origins are allowed to send CORS requests (this is defined by the origin HTTP request header). The following configuration in elasticsearch.yml will allow ALL origins but is considered insecure:

http.cors.allow-origin: "*"

For more detailed information on Elasticsearch configuration options refer to: Elasticsearch Configuration Reference

Enabling CORS with a proxy

As an alternative to enabling CORS through the Elasticsearch configuration file, you can setup a proxy in front of Elasticsearch that will set the proper HTTP response headers.

As an example, in AWS - here's a link that describes how to setup an API gateway that sends CORS headers: http://enable-cors.org/server_awsapigateway.html

An instance of an API gateway (with CORS enabled) is used that forwards requests to your Elasticsearch instance.

The Elasticsearch URL used in the Tableau connector configuration should be the URL of your API Gateway.

Building and Running

Pre-requisites

Install grunt:

npm install -g grunt

Install bower:

npm install -g bower

Install dependencies

Run the following from the command line:

npm install
bower install

Creating distribution suitable for deploying to Tableau Server

From the command line execute:

grunt build:dist

This will package the connector files in the dist folder, combining javascript and CSS into single files.

Building and running from local development environment

You can build, watch sources for changes, and run the application at the command line with grunt with:

grunt

This will watch all sub-directories for changes and reload the application if anything changes. Running this app will simply host all the connector resources but when requested stand alone will not do anything useful. You should either use the connector within the Web Data Connector SDK test harness, or use the connector from Tableau Desktop or Server.

Note that internally there are tasks that run the build:dev target, to perform HTML templating, and copy all source files to the public/ source folder where the NodeJS server will serve static resources from

Make note of the URL that the connector app is running on, e.g.:

Elasticsearch Tableau Web Data connector server listening at http://0.0.0.0:3000

Simply choose the 'Web Data Connector' as your data source from within Tableau Desktop, or use the Web Connector SDK and enter the URL..

Running using Docker

A Dockerfile is supplied in docker/Dockerfile that will build an image that creates a development build of the latest source from Github, and runs the node server.

You can build an image from the root of the project:

docker build docker -t <name of tag>

and can then start a container, which will map the server to the host's port 3000 from this image with:

docker run <name of tag> -p 3000:3000

Running as a Windows service

For convenience, the connector comes with winser to install the connector web server as a Windows service.

To install as a Windows service:

npm run-script install-service

This will install a service named elasticsearch-tableau-connector. Open the Windows service manager (services.msc) to start the service.

To uninstall the service:

npm run-script uninstall-service

Loading the Connector UI outside Tableau

The connector UI can be loaded from a web browser (outside of Tableau Desktop). Simply enter the URL of the connector (defaults to http://localhost:3000/elasticsearch-connector.html if running the project locally).

The 'Submit' button will not be displayed, but you can still use the 'Preview' feature of the connector.

Importing into Tableau Server

Execute the build for this project from the command line:

grunt build:dist

For each file in the dist/ folder, import into Tableau Server by:

tabadmin import_webdataconnector elasticsearch-connector.html
tabadmin import_webdataconnector elasticsearch-connector.min.css
tabadmin import_webdataconnector elasticsearch-connector.min.js

Get the URL of the elasticsearch-connector.html on the Tableau Server by executing:

tabadmin list_webdataconnectors --urls

And from Tableau go to 'Web Data Connector' and enter the URL of the connector:

http://<your tableau server>/webdataconnectors/elasticsearch-connector.html

Using with the Web Data Connector SDK

If you are running this web app locally, and testing from the Tableau Web Data Connector SDK, simply enter:

http://localhost:3000/elasticsearch-connector.html

into the Web Connector URL input field in the SDK's form.

From there you should see this connector's UI:

Connector UI when in Aggregation mode:

Connector UI after fetching preview data:

Using the Connector UI

The Elasticsearch connector UI includes the following fields:

Field NameData TypeDescription
Connection NameStringName of the data source connection displayed in the Tableau workbook
Elasticsearch URLString[Required] URL of the Elasticsearch cluster
Use HTTP Basic authenticationBoolean[Required] Indicates if the Elasticsearch cluster requires HTTP Basic Auth
UsernameStringIf 'Use HTTP Basic Auth' is checked, this is the user name
PasswordStringIf 'Use HTTP Basic Auth' is checked, this is the password
Index nameString[Required] Name of the index in the Elasticsearch cluster
Index Filter for Type SelectionStringIf the index selected is an alias, this selection is required to choose the index to filter types by. Only types from this selection will be available in the 'Type' selection
Use fields for type from all indexes?BooleanIf selected, then fields in the extract will be a union of all fields of the selected type across all indexes of the alias. Note this only applies when selecting an index alias. Defaults to false
TypeString[Required] Name of the type in the Elasticsearch cluster to query
Override Field DefaultsBooleanIf selected, then additional options to override default handling of fields is available
Parse date fields in local time?BooleanIf selected then all date or timestamp fields will be parsed in local time, the default (unselected) will parse as UTC
Improve Field Names?BooleanIf selected, then additional logic will be applied to improve field names in similar ways to the following : Tableau help article. Default is false.
Result ModeOptionOption to retrieve search results from Elasticsearch (Search Result Mode) or from a query using aggregation (Aggregation Mode)
Use custom query?BooleanIf true, indicates if the extract should use a custom query against Elasticsearch in search result mode, if false extract will be a 'match all'
QueryStringIf Use custom query? is true, this will be the JSON request payload sent to Elasticsearch in search result mode. from, and size will be overwritten if supplied. Refer to Elasticsearch Query DSL for a reference on writing a query
Use Incremental RefreshBooleanIf checked, then Tableau can fetch data using incremental refresh mode
Incremental Refresh ColumnStringName of the column to use for incremental refreshes. Should be a date, time or integer column
Batch size of per request to ElasticsearchIntegerNumber of rows to retrieve at once, defaults to 10, should probably be 1000+
Total limit on number of rows to syncIntegerLimit of rows to include in extract, defaults to 100, but generally should be left blank to indicate that all matching rows should be included
Use custom query? (aggregation mode)BooleanIf true, indicates the data extract should use a custom query that includes an aggregation request
Custom queryStringJSON payload sent in the request for Elasticsearch, must include aggregations or aggs element for Terms, Range, Date Range or Date Histogram
Filter data included in aggregationsBooleanIf checked, then you can enter a filter that will be used against data used in an aggregation request
FilterStringUses Lucene Query String Syntax to define a filter applied against aggregation data
MetricsMetricOne or more metrics to calculate for the aggregation results. Valid options are Count, Min, Max, Sum, Average, Stats, and Extended Stats. Refer to 'Metrics' section
BucketsBucketBucket to aggregate results to and calculate metrics for, or multiple levels of child buckets. See buckets for more information

Metrics

Supported metrics:

Buckets

Preview

The connector supports requesting data for Elasticsearch from the UI to preview the data that will be created in the data extract in Tableau. The preview button will send this request to Elasticsearch based on the current configuration and populate a table at the bottom of the view. This feature is useful for debugging to make sure any custom queries and other configuration returns a valid response.

Note - it is recommended to set a small limit if in 'Search result mode' to limit the amount of data returned

Submit

The submit button will save the configuration for the data extract with Tableau and continue the process of creating the extract.

Elasticsearch Field Renaming

Tableau only supports field names with alphanumeric characters and underscores. The connector will replace non-supported characters with '_' underscore characters.

e.g.:

Elasticsearch field nameResulting 'safe' Tableau field name
field_namefield_name
@timestamp_timestamp
name.firstname_first
car.@namecar__name

Note: Field names available to select from the Connector UI will be the Elasticsearch field names, but data in Preview Mode and what is actually provided to Tableau will be converted to the safe names

Note: Additional logic will be applied to rename fields according to: Tableau help articlewhen 'Improve Field Names' has been selected under 'Override Field Defaults'

Handling Elasticsearch Data Types

object

For types that include mapping with objects (fields with their set of properties), a concatenated field name will be created. For the following mapping:

{
    "person": {
        "properties": {
            "firstName": {
                "type": "string"
            },
            "lastName": {
                "type": "string"
            },
            "address": {
                "properties": {
                    "street": {
                        "type": "string"
                    },
                    "city": {
                        "type": "string"
                    },
                    "zip": {
                        "type": "string"
                    }
                }
            }
        }
    }
}

Will create the following fields:

geo_point

For geo_point fields in Elasticsearch, this connector will create two separate Tableau fields by parsing the lat, lon value:

Incremental Refresh

The connector supports Tableau's incremental refresh feature in 'Search Result' mode. This can be used to extract large datasets from Elasticsearch that can be incrementally imported into Tableau.

Your Elasticsearch type should have a date, time or integer field that is used to query for incremental data. The last value for this column is stored and used on subsequent extracts as the starting value. E.g., if the last value seen for a field @timestamp was 1/1/2000 00:00:00 then the next time an incremental extract is processed, the query to Elasticsearch will filter on the @timestamp field for values greater than 1/1/2000 00:00:00.

Generally the value should be unique and be automatically incremented as new data is added to Elasticsearch (why a timestamp or auto incrementing sequence number are good choices).

For more information refer to:

Sponsorship

DialogTech Logo

This project has been made possible in part by support from DialogTech