Urban Institute Advances Mission, from Security to Speed to Capacity, with the Cloud

The Urban Institute is a Washington, D.C.-based policy research organization dedicated to developing evidence-based insights that improve decision making, strengthen communities, and make people’s lives better.

As a leader in data-driven policy research, the Urban Institute was looking for a solution that would enable it to run sophisticated data analytics and microsimulation models with flexibility, speed, and accuracy.

The organization was running complex, data-driven microsimulations developed over years, and wanted to evolve its processing platform while retaining select legacy code. They also hoped to improve information sharing, lower costs for storage and processing, and be able to ramp up quickly.

The Urban Institute chose a cloud-first strategy starting with Amazon Web Services (AWS). Urban tested multiple applications to determine how AWS could work across the organization, from data research to enterprise systems.

 “The more the team started to learn about the capabilities of the cloud, the more AWS became a natural fit to help advance our research mission, from security to speed to capacity,” said Khuloud Odeh, VP for IT and CIO, Urban Institute.

Urban Institute piloted a microsimulation model cloud migration with its Tax Policy Center, designing a new architecture in the cloud that did not require many changes to the original model source code. Using AWS, the team was able to spin up more machines faster and conduct high-performance parallel processing. Urban was able to run the model using multiple scenarios at the same time and get all of the results at once. This group served as internal advocates to get other teams on board and migrate other models to the cloud.

In addition to increasing performance, the cloud opened doors for collaboration. By using AWS, researchers can access their work at home and share files to increase collaboration across teams. For example, Urban’s Spark for Social Science project—hosted on GitHub—delivers powerful and elastic Amazon Elastic MapReduce (EMR) Spark clusters to researchers and data analysts at Urban and the public social science community with minimal setup time. It takes only 12-15 minutes to spin up a cluster.

Using the AWS Cloud, researchers are able to experiment and learn with new research possibilities and fewer constraints. They are now thinking of new research questions they thought they would never have the infrastructure to answer. Urban researchers are thinking big, with no server capacity limitations.