Planet: Imaging a dynamic, ever-changing Earth

About Planet Labs, Inc.

Planet is an integrated aerospace and data analytics company that operates history's largest fleet of Earth-imaging satellites, collecting a massive amount of information about our changing planet.

Industries: Technology
Location: United States

To deliver next-generation geospatial services to users, Planet moved all of its satellite imagery and image processing to Google Cloud Platform as part of a multi-cloud strategy.

Google Cloud Results

  • Processes fresh satellite images of the Earth’s landmass every 24 hours
  • Supports a 6x capacity increase
  • Reduces data processing costs up to 60%

Collecting data from 190+ satellites

The Earth is changing fast. The ability to visualize how it’s changing can lead to new discoveries and help decision-makers in commercial and government sectors take faster action to address many of the world’s problems, from providing business intelligence to corporations to tracking famine and deforestation. It can also help private companies gain valuable insights into agriculture and commerce.

To this end, Planet Labs, Inc. (Planet) has launched the largest constellation of satellites in human history to capture daily images of the entire Earth’s landmass on a daily basis. Today, Planet has more functional satellites in orbit than any government. Its focus is to democratize access to satellite imagery so that a new ecosystem of users can take action with the data to make a difference in commercial and environmental sectors. To make its venture financially possible, Planet uses lean, low-cost electronics combined with cloud computing.

“The resources required to handle our satellite data are immense. Completing Mission 1 meant processing 7 to 10 terabytes of data daily. To keep our team focused on what makes Planet unique, we needed a cloud provider that could offer proven reliability on a planetary scale.”

Troy Toman, Director of Engineering, Planet

Planet began with just 4 satellites in 2013 but quickly scaled up, deploying 88 satellites in a single record-setting launch in February 2017. The company set its sights on what it called Mission 1—imaging the entire Earth’s landmass every 24 hours. Once the satellites started operating at production rates, they began transmitting terabytes of data every day to ground stations all over Earth. Planet needed a compute and storage platform that could handle the load, and scale to petabytes of data and tens of thousands of virtual machines (VMs), to accommodate new satellite launches every three to four months.

“The resources required to handle our satellite data are immense,” says Troy Toman, Director of Engineering for Planet. “Completing Mission 1 meant processing 7 to 10 terabytes of data daily. To keep our team focused on what makes Planet unique, we needed a cloud provider that could offer proven reliability on a planetary scale.”

Mission accomplished

As part of a multi-cloud strategy, Planet decided to move all its satellite data processing to Google Cloud Platform and use Google Cloud Storage to host its image archive. Planet selected Google for several key advantages: cost efficiency and predictability, custom machine types, and future access to industry-leading machine learning technology. It also wanted the ability to use Google Earth Engine datasets on Google Cloud Storage, such as radar data from Sentinel-1, to give customers access to a variety of datasets. Google’s high-speed network and worldwide connectivity help Planet collect data from its ground stations around the world.

“Only Google Cloud Platform could provide the core storage and compute services we needed to manage our multi-terabyte daily satellite data processing pipeline and make it scalable and affordable,” says Troy. “Google’s focus on multi-region services fits our global data processing model well, giving us many options around the globe.”

Planet worked with Google Cloud consultants and a Google Cloud Technical Account Manager to help ensure a smooth migration to Google Cloud Platform. The Google professional services team reviewed Planet’s migration plan and helped determine the best way to move the company’s large storage resources to Google Cloud Platform. They also helped Planet understand the range of tools and APIs available to help with the transition, and facilitated discussions with product teams to answer Planet’s questions.

“Our three-month migration to Google Cloud Platform was nearly invisible,” says Troy. “A lot of the credit goes to the Google professional services team. They were hugely helpful in answering our questions and moving our 7PB of satellite data, and probably cut our migration time in half by becoming part of our team.”

“We find the Google Cloud Platform pricing model predictable and easy to understand, reducing our data processing costs by up to 60%. We often run more than 40,000 preemptible VMs concurrently to match our incoming queue for image processing, so we’re saving money that we can redirect to other areas.”

Troy Toman, Director of Engineering, Planet

With the ability to process data as quickly as satellites can transmit it, Planet soon declared Mission 1 a success. It has since added more satellites, bringing its total count to over 190. As a result of having more satellites in orbit, data is growing even faster. Planet now adds as much as 7TB a day, and post processes up to 12TB a day.

“Moving our satellite data processing to Google Cloud Platform was critical to achieving our Mission 1 goal of imaging the Earth’s entire landmass every day,” says Troy. “We’ve built a queryable Earth that is always updated and fresh—and that opens up vast learning opportunities for scientists, governments, and humanitarian causes.”

Reducing data processing costs by up to 60%

To keep costs down, Planet uses preemptible virtual machines (VMs) on Google Compute Engine, using affordable, short-lived compute instances for its batch processing. Planet can easily configure instances that are the best fit, with the exact amount of memory, disk capacity, and CPU cores that are needed.

“We find the Google Cloud Platform pricing model predictable and easy to understand, reducing our data processing costs up to 60%,” says Troy. “We often run more than 40,000 preemptible VMs concurrently to match our incoming queue for image processing, so we’re saving money that we can redirect to other areas.”

Planet uses Google Virtual Private Cloud (VPC) to share networks across projects and Google Cloud Load Balancing to run services behind a private load balancing IP address accessible only to internal VPC instances.

“Google VPC and Google Cloud Load Balancing on Google Cloud Platform allow us great flexibility while maintaining a high degree of privacy and security,” says Troy. “We’ve been using both since they were in alpha.”

Tracking billions of objects

As Planet’s satellite image archive grows, so does the metadata surrounding the images. To scale data tracking without adding headcount to its team, Planet re-architected its primary storage application to leverage Google Cloud Bigtable, a managed database service for big data, instead of self-managed PostgreSQL databases. It also uses Google Cloud Datastore and Google BigQuery to store configuration data and analyze CPU, memory, and disk usage data for performance tuning while scanning millions of images a day.

“Our satellite image store grew by 600% in six months and our data tracking system has processed more than two billion objects,” says Troy. “Without managed services like Google Cloud Bigtable and Google Cloud Datastore, it would be very difficult for us to scale without dedicating at least two full-time employees.”

“Since moving our image processing to Google Cloud Platform, we’ve increased the number of satellite images we’re processing sixfold. Google Cloud Platform allowed us to scale in an elegant, unobtrusive fashion during a period of tremendous growth.”

Troy Toman, Director of Engineering, Planet

Making an impact from above

For now, Planet is unique in the market, and its technology is ushering in a new era of possibilities. With fast insights into land use changes, agriculture companies can maximize yield by making more intelligent decisions about crop health and weather patterns. Countries can better respond to disasters, monitor deforestation and take steps to slow the pace, and keep track of activities along their borders.

“Since moving our image processing to Google Cloud Platform, we’ve increased the number of satellite images we’re processing sixfold,” says Troy. “Google Cloud Platform allowed us to scale in an elegant, unobtrusive fashion during a period of tremendous growth.”

Collaborating with Google will help Planet keep its competitive edge while enabling humanitarian, scientific, and commercial advances that might not otherwise be possible. “Philosophically, we are aligned with Google’s focus on services, leveraging open source projects like Kubernetes and TensorFlow,” says Troy. “As our data and compute infrastructure grow, we’re excited to work with Google Cloud to deliver next-generation geospatial services.”

About Planet Labs, Inc.

Planet is an integrated aerospace and data analytics company that operates history's largest fleet of Earth-imaging satellites, collecting a massive amount of information about our changing planet.

Industries: Technology
Location: United States
Google Cloud Platform logo

12 Months FREE TRIAL

Try Kubernetes Engine, BigQuery, and other Cloud Platform products with $300 in free credit and 12 months.

TRY IT FREE
Google Cloud Platform logo

12 Months FREE TRIAL

Try Kubernetes Engine, BigQuery, and other Cloud Platform products with $300 in free credit and 12 months.

TRY IT FREE