LookML usage

The LookML Library for MDE provides abstract core files, and packaged views, explores and dashboards to help you get started with building your own LookML models on top of the MDE data in BigQuery.

Before you begin

To complete this guide, you need the following prerequisites:

  • You configured a new Looker project using the MDE Project Template as described in the setup guide.
  • You have an active MDE deployment.
  • Optional. You applied the default MDE configuration package. While you can use the MDE LookML Library to create custom views, the default LookML package requires the default MDE configuration package is applied.

Exploring library contents

After the project is allowed, you can explore the contents of the library under the imported_projects/mde_lookml_lib directory in the Looker development console. The library provides content in two directories:

  • mde_abstract_core
  • mde_packages

Abstract core directory

The mde_abstract_core directory provides base views that you can use to model MDE metadata buckets as dimensions and MDE record tables as fact tables in Looker. The base views save you from needing define the shared fields across these views.

The library provides a dim_base.view view for building dimensions, which you can find in imported_projects/mde_lookml_lib/mde_abstract_core/views/dimension/dim_base.view.lkml. The following example shows how you can extend the base view to define a dimension called device:


include: "//mde_lookml_lib/mde_abstract_core/views/dimension/*.view"

view: dim_source {

   derived_table: {
      sql: SELECT * FROM `mde_dimension.metadata-store` where bucket_number = BUCKET_NUMBER ;;
   }

   view_label: "DIM - Device"
   extends: [ dim_base ]

   dimension: device_name {
      type: string
      sql: JSON_VALUE(${TABLE}.instance,"$.device_name") ;;
   }

}

And, similarly, the library provides base views per MDE archetype for modeling record tables in BigQuery as fact tables. You can find the base views under imported_projects/mde_lookml_lib/mde_abstract_core/views/fact/. You can extend the archetype base views to define custom fact views in Looker. The following example shows how you can define a fact view called sensor_readings:


include: "//mde_lookml_lib/mde_abstract_core/views/fact/fct_numeric_archetype.view"

view: fct_sensor_readings {
   extends: [ fct_numeric_archetype ]
   label: "FCT - Sensor Readings"
   sql_table_name: `mde_data.sensor-readings` ;;
}

Packages

The mde_packages directory contains prebuilt LookML packages for MDE. These packages provide use-case specific explores, views and dashboards to get you started quickly.

The default Looker configuration package works in conjunction with the default MDE configuration package. For more information, see the Default configuration package guide. The default Looker configuration package provides a fact table per record table linked to an MDE type, as well tag and source dimensions, which correspond to the tag and source metadata buckets in MDE, respectively.

If you used the MDE Project Template to bootstrap a new MDE project, the provided Looker model automatically imports the views and explores from the default package:

connection: "@{looker_connection}"

########## DEFAULT MDE EXPLORES ##########
include: "//mde_lookml_lib/mde_packages/default/explores/default.explore.lkml"

############ CUSTOM VIEWS ################
include: "/views/dimension/*.view"
include: "/views/fact/*.view"

Once your project is set up, you can use the default default explores in Looker to analyze data.

You can also use other configuration packages included in the library. See the following section for further instructions and prerequisites to use these packages in the library.

Extend views provided by the configuration packages in the library

You can extend the fact and dimension views included in the library by using the extends syntax. For example, in your project model file you can extend the dim_tag by defining additional dimension fields on the deviceMetadata object, as the following script shows:

view: +dim_tag {
dimension: device_type {
type: string
sql: JSON_VALUE(${TABLE}.instance,"$.deviceMetadata.deviceType") ;;
}
dimension: location {
type: string
sql: JSON_VALUE(${TABLE}.instance,"$.deviceMetadata.location") ;;
}
}

Dashboards access is defined in the default configuration package

The LookML model defined in the MDE Project Template imports all the dashboards from the default package in MDE LookML Library. To access any of these dashboards, open the following link:

https://LOOK_HOST/dashboards/mde::DASHBOARD_ID

Replace the following:

  • LOOK_HOST with the hostname of your Looker instance.
  • DASHBOARD_ID with the ID of the dashboard you want to open. You can look up the dashboard ID, by opening the relevant .dashboard file in the imported_projects/mde_lookml_lib/mde_packages/default/dashboards/ directory and copying the value of the dashboard property.