Create a Streaming Data Pipeline on GCP with Cloud Pub/Sub, Dataflow, and BigQuery

45 minutes
  • 8 Learning Objectives

About this Hands-on Lab

This lab will simulate live highway sensor data which will be published to a Cloud Pub/Sub topic. Then, a Cloud Dataflow streaming pipeline will subscribe to it. The pipeline will take the streaming sensor data, transform it, and insert it into a BigQuery table. We will then view the streaming inserts in BigQuery while they are in progress, and attempt to gain some useful insights from the streaming data.

Learning Objectives

Successfully complete this lab by achieving the following learning objectives:

Prepare Your Environment

Enable pub/sub and dataflow APIs:

gcloud services enable
gcloud services enable

Create a Cloud Storage bucket for Dataflow staging:

gsutil mb gs://$DEVSHELL_PROJECT_ID

Download the GitHub repository used for lab resources:

cd ~
git clone
Create Pub/Sub Topic
gcloud pubsub topics create sandiego
Create a BigQuery Dataset to Stream Data Into

Create a BigQuery dataset to stream data into:

bq mk --dataset $DEVSHELL_PROJECT_ID:demos

The table will be named average_speeds. We do not create the table, but Dataflow will create it within the dataset for us.

View the Dataflow Template

We will not be interacting with the template directly. We will be using a script that will install the Java environment and execute the template as a Dataflow job:

vim googledataengineer/courses/streaming/process/sandiego/src/main/java/com/google/cloud/training/dataanalyst/sandiego/
Create the Dataflow Streaming Job

Go to the Dataflow job script directory:

cd ~/googledataengineer/courses/streaming/process/sandiego 

Execute the script that creates the Dataflow streaming job, and subscribe to the Pub/Sub topic.

This script passes along the Project ID, staging bucket (also the Project ID), and the name of the Java template to use:


When complete, the streaming job will be subscribed to our Pub/Sub topic, and waiting for streaming input from our simulated sensor data.

Publish Simulated Traffic Sensor Data to Pub/Sub via a Python Script and Pre-Created Dataset

Browse to the Python script directory:

cd ~/googledataengineer/courses/streaming/publish

Install any requirements for the Python script:

sudo pip install -U google-cloud-pubsub

Download the simulated sensor data:

gsutil cp gs://acg-gcloud-course-resources/sandiego/sensor_obs2008.csv.gz .

Execute the Python script to publish simulated streaming data to Pub/Sub:

./ --speedFactor=60 --project=$DEVSHELL_PROJECT_ID
View the Streamed Data in BigQuery

In BigQuery, execute the following query to view the current streamed data, both in the table and in the streaming buffer:

FROM `demos.average_speeds`

Notice the total count of records at the bottom. Wait about a minute and run the same query again (be sure to uncheck use cached results in query options) and notice that the number has increased.

Use Aggregated Queries to Gain Insights

Let’s get some use out of this data. If we wanted to forecast some necessary road maintenance, we would need to know which lanes have the most traffic, to know which ones will require resurfacing first.

Enter the following query to view which lanes have the most sensor counts:

SELECT lane, count(lane) as total
FROM `demos.average_speeds`

We can also view which lanes have the highest average speeds:

SELECT lane, avg(speed) as average_speed
FROM `demos.average_speeds`
ORDER BY average_speed DESC

Additional Resources

Many data engineer scenarios on GCP involve a multi-step streaming data pipeline from ingestion, to processing, to storage/analysis. In this lab, we will create a simulated end to end streaming pipeline of all steps, which will finish in analyzing captured streaming data for insights.

Be sure to launch the lab in your browser's incognito (or other private browsing) mode to avoid cached login issue.

What are Hands-on Labs

Hands-on Labs are real environments created by industry experts to help you learn. These environments help you gain knowledge and experience, practice without compromising your system, test without risk, destroy without fear, and let you learn from your mistakes. Hands-on Labs: practice your skills before delivering in the real world.

Sign In
Welcome Back!

Psst…this one if you’ve been moved to ACG!

Get Started
Who’s going to be learning?