Build a GraphQL API with BigQuery Datasets and Hasura

Google Cloud BigQuery is a key service that helps you create a Data Warehouse that provides the scale and ease of querying large data sets. Let’s say that you have standardized on using BigQuery and have set up data pipelines to maintain the datasets. The next question would be to determine how best to make this data available to applications. APIs are often the way forward for this and what I was looking to experiment with is to consider a service that helps me create an API around my data sources (BigQuery in this case) and do it easily.

In this blog post, we shall see how to use Hasura, an open-source solution, that helped me create an API around my BigQuery dataset.

The reason to go with Hasura is the ease with which you can expose your domain data via an API. Hasura supports a variety of data sources including BigQuery, Google Cloud SQL and AlloyDB. You control the model, relationships, validation and authorization logic through metadata configuration. Hasura consumes this metadata to generate your GraphQL and REST APIs. It’s a low-code data to API experience, without compromising any of the flexibility, performance or security you need in your data API.

While Hasura is open-source, it also has fully managed offerings on various cloud providers including Google Cloud.

Pre-requisites

You need to have a Google Cloud Project. Do note down the Project Id of the project since we will need to use that later in the configuration in Hasura.

BigQuery dataset – Google Trends dataset

Our final goal is to have a GraphQL API around our BigQuery dataset. So what we need to have in place is a BigQuery dataset. I have chosen the Google Trends database that is made available in the Public Datasets program in BigQuery. This is an interesting dataset that makes available (both US and Internationally), the top 25 overall or top 25 rising queries from Google Trends from the past 30 days.

I have created a sample dataset in BigQuery in my Google Cloud project named ‘google_trends’ and have copied the dataset and the tables from the bigquery-public-data dataset. The tables are shown below: