Verily leverages SADA’s Google Cloud expertise to migrate 48 petabytes of data and applications

According to an ancient Chinese proverb, “A journey of a thousand miles begins with a single step.” For Verily, a health data platform and technology company, born as a Google X moonshot, their trip through time and space had taken them to the point where they were ready to separate from the Google booster rocket and become an independent company within Alphabet.

“Verily was founded to solve audacious problems in healthcare, with the goal of ultimately standing outside of Google itself,” says Stas Zvinyatskovsky, Distinguished Engineer and former Head of Engineering Platform at Verily. “Verily was born in 2016, since which time we’ve made many health technology advancements.” 

Due to the company’s origins, Verily ran their data and applications within Google’s internal infrastructure. To operate more independently, Verily prepared to separate their technology infrastructure from Google, and they needed to migrate their data assets and applications to the public Google Cloud environment. Given the volume and complexity of Verily’s data and applications, this was a daunting task, even for the bravest infrastructure engineer.

Business challenge

Meeting the objectives of migrating from internal Google systems

Complicating the challenge of moving off Google’s internal infrastructure, Verily had an ambitious timeline to complete the transfer of 48 petabytes of data and applications. (By comparison, the U.S. Library of Congress contains 184 petabytes of information.) Other key challenges included:

• Complexity and volume of data accumulated over nearly a decade
• Need for a centralized approach to data migration to ensure efficiency and handle knowledge gaps within individual teams

“Getting off Google’s infrastructure, moving our data and apps, and setting up our own environment were all key objectives,” says Zvinyatskovsky. 

Moving all products, healthcare data, and sandbox projects

The ultimate goal of Verily’s migration was to move all of the 48 petabytes of project data and applications that had been developed during their tenure in the Google infrastructure, including the data generated from clinical studies and Verily’s care solutions. Not only did they need to transfer all product projects and related data, but also anything that had been created in a sandbox environment that hadn’t moved to production. 

“In addition to production projects and resulting research and care data, there were experimental projects, where people were able to play around in a sandbox and see exactly what they could do,” says Zvinyatskovsky.

Solution

To solve their business challenges with migrating from the Google corporate infrastructure to one based on a separate Google Cloud instance, Verily wanted to collaborate with a company steeped in Google Cloud technology. Verily chose to collaborate with SADA, an Insight company and multiple-time Google Cloud Partner of the Year, which is stacked with former Google employees who understand Google’s systems and have deep Google Cloud expertise. 

These SADA-Google Cloud experts gave Zvinyatskovsky and Verily the confidence to move forward with their aggressive timeline to migrate 48 petabytes of data and applications. “We valued SADA for their Google Cloud expertise and ability to quickly stand up this migration,” says Zvinyatskovsky. “SADA’s experienced help kept our engineering experts focused on their day jobs and not on the migration, which SADA was well positioned to take on.”

Planning ahead for the migration

“There was significant work spent on preparation, and it was helpful in that way for understanding and access controls,” says Zvinyatskovsky.

In preparation for the migration, SADA and Verily collaborated closely, with weekly status updates, monthly executive steering committee meetings, and regular communication channels to ensure alignment and address any issues that arose.

“The goal of our meetings and constant communication was to create a centralized project,” says Zvinyatskovsky. “We wanted to hone in and make sense of things by cataloging all the necessary data that we’re moving and organize it in a centralized manner versus bringing the data over en masse and not understanding it.”

Breaking up the migration into manageable pieces

In implementing a phased approach, SADA and Verily ended up dividing the data into 19 batches to mitigate risk and ensure smooth execution. SADA’s solution also involved: 

• Utilizing Cloud Composer to orchestrate and automate the migration, ensure accuracy, and provide real-time visibility into migration status
• Leveraging dynamic and static directed acyclic graphs for the execution of asset inventory, metadata collection, migration automation, and state management
• Transferring data via Storage Transfer Service for bucket migration and migrating datasets via BigQuery Transfer Service
• Establishing BigQuery tables for metadata, state management, and real-time reporting 
• Creating real-time dashboards accessible by stakeholders for migration tracking

Impact

As a result of collaborating with SADA, Verily was able to complete their migration two months ahead of schedule and under budget, with 48 petabytes of data, 1.8 million BigQuery tables, and over 5 billion objects successfully transitioned. This improved Verily’s research and development capabilities by providing data scientists with access to reliable and complete data sets for training models. 

“The migration helped our data science team by knowing that all their data is available, so that they could train on it without any interruptions,” says Zvinyatskovsky. “Our data scientists are great at training their models, and SADA’s Google Cloud storage and migration experts ensured that we had moved the data for them, so they’re confident that their training data is all there. That has really helped out our R&D folks.” 

Driving innovation in healthcare and life sciences

One of the significant outcomes from the migration has been the ability of Verily to supercharge innovation to deliver new care and research solutions to market. Already, they are planning for the launch in 2026 of Lightpath, Verily’s next-generation chronic care management solution. 

“Lightpath will feature an AI-enhanced digital health coach and uses data to stratify members and put them into the right level of program that’s appropriate for their care needs,” says Zvinyatskovsky. “It’ll use data as a continuous feedback loop to continue optimizing and providing recommendations for patients. It’s very much dependent on our new data platform and having all our data in the new Google Cloud environment.”

Verily will productize the platform powering Lightpath so their customers can make their own breakthroughs in healthcare and life sciences. “Verily will offer the Verily platform to customers to build their own apps and solutions,” says Zvinyatskovsky. “Having built this infrastructure and an ecosystem and tools to analyze the data, it has become the core of the business.”

Achieving key benefits 

Overall, Verily was able to obtain key benefits due to their collaboration with SADA on their migration from Google corporate infrastructure to their own Google Cloud environment. These benefits include:

• Enhanced data accessibility and reliability for research and development
• Increased efficiency and visibility through centralized data management
• Cost savings due to automation and optimization
• Improved internal capabilities and expertise through collaboration with SADA

The successful migration empowered Verily to operate more independently, optimize their data management practices, and accelerate innovation in the healthcare and life sciences sectors using Google Cloud.