Skip to main content

Run GOAT

What an install is and who looks after it.

This guide is for the people who install GOAT and look after it: cloud engineers and administrators at a company that runs GOAT. It covers what an install is, what it creates in your Google Cloud project, and the tasks of running it.

What an install is​

An install is one copy of GOAT in one of your Google Cloud projects. Everything it needs runs in that project, and you pay Google directly for what it uses. There is no GOAT-hosted service in between, and an install does not update itself or report back to anyone.

Each install has its own address, such as https://analytics.example.com, its own operators, its own datasets and dashboards, and its own customers.

What it creates in your project​

PartWhat it is for
Cloud Run serviceRuns GOAT: the app, the API, and the embed script. One instance stays warm.
Firestore databaseHolds what operators build: datasets' settings, collections, charts, dashboards, tenants, and access.
BigQuery datasetHolds GOAT's managed copies of your data.
Cloud Storage bucketHolds the files operators upload.
Identity PlatformSigns in operators, by password with an optional second factor, or with Google.
Secret Manager secretHolds the keys that sign viewer tokens.
A second BigQuery dataset and a log sinkKeep a record of each query, for usage and cost figures.
A service accountThe identity GOAT runs as, with only the access listed in Architecture on Google Cloud.

GOAT gets no access to your own data when it is installed. You grant it, dataset by dataset and bucket by bucket. See Give GOAT access to your data.

The tasks​

TaskPage
Install GOATInstall GOAT
Decide who signs in to buildOperators and sign-in
Give backends their API keysAPI keys
Let GOAT read your dataGive GOAT access to your data
Know and cap what it costsCosts
Take a new releaseUpgrades

For how the parts work together, see Under the hood.

GOAT documentation

Type to search.