Mississippi Department of Employment Security Cuts Infrastructure Costs by 72% on AWS

In 2012, the Mississippi Department of Employment Security (MDES) had a vision of a consortium between several states to operate a multitenant unemployment insurance (UI) system. Knowing its data center couldn’t support such a goal, the government agency decided to migrate the system to the cloud. “We knew the final destination for the system had to be the cloud because it’s a shared system,” says Mohammed Jalaluddin, chief technology officer of MDES. “We couldn’t imagine any state hosting it alone.”

After initially trying another cloud provider, MDES turned to Amazon Web Services (AWS), migrating the system and refactoring it as a multitenant system that can be shared with other states. The result, ReEmployUSA, became the first UI system shared by multiple US states to be hosted in the cloud. By hosting ReEmployUSA on AWS, MDES has achieved high security and scalability, reduced the system’s operating costs, and helped other states take advantage of the system.

MDES collects unemployment taxes and provides an unemployment benefits program for the state. It also works with approximately 55,000 employers in Mississippi to connect them with qualified employees through its workforce services. The agency developed the idea of the consortium, which centered on the use of a shared cloud system, so that partnering states could optimize the infrastructure for the long term and reduce costs. MDES received funding from the US Department of Labor in 2012 to take the lead in developing this consortium, and now ReEmployUSA includes Mississippi, Connecticut, and Maine.

At the time, there were no other shared UI systems on the cloud, and many states were dubious of the shared system’s viability. Security was the most prominent concern: MDES had to convince other states that a shared cloud system would be at least as secure as an on-premises system. MDES’s on-premises data center couldn’t offer the scalability and high availability that the consortium would require and couldn’t meet the levels of security mandated by the Federal Risk and Authorization Management Program. The agency also had to negotiate the security agreements with the Internal Revenue Service and the Social Security Administration.

MDES chose a cloud provider in 2016, and ReEmployUSA launched in 2017. But although MDES could scale down the infrastructure, its costs didn’t scale down accordingly. It evaluated several different cloud solutions and ultimately selected AWS due to scalability as well as the ease of use of its operating system. MDES decided to migrate ReEmployUSA fully to AWS in 2019. On AWS, MDES could scale down and only pay for the resources it consumed, which would cut costs significantly. The agency also preferred AWS for its ease of use in migrating from a monolithic application to a microservices-based one. MDES could accomplish this using Amazon Elastic Kubernetes Service (Amazon EKS)—a managed container service for running and scaling Kubernetes applications—to access highly available and secure Kubernetes clusters and automate key tasks such as patching, node provisioning, and updating.

For its migration and modernization effort, MDES engaged AWS Partner Tata Consultancy Services, an information technology services, consulting, and business solutions organization that had been working with the agency on its UI and tax systems since 2005. With help from Tata Consultancy Services, MDES completed the migration from its previous vendor to AWS in less than 8 hours in June 2020, when the workload was at its peak because of the COVID-19 pandemic. “Our migration to AWS was so seamless and simple that it was a nonevent,” says Jalaluddin. MDES pays $17,000 a month for its AWS infrastructure—about 72 percent less than it paid the previous cloud provider. “By buying Red Hat Linux licenses through AWS for nonproduction, we only pay for what we use,” says Jalaluddin.

Once ReEmployUSA was set up, MDES began helping other states migrate to the multitenant system. It started with Connecticut because the state needed to begin hosting a temporary COVID-19 pandemic unemployment system quickly to accommodate the provisions of the Coronavirus Aid, Relief, and Economic Security Act.

As part of the preonboarding process, MDES uses gap analysis to identify how it can adapt the system to suit a new state. Since the COVID-19 pandemic began, MDES has paused onboarding other states, but it expects to begin again in 2022. However, Missouri, Wyoming, and New York have also used the system as a foundation for their own versions: “We have given our system, free of cost, to other states that want to use it as a base and do their own development,” says Jalaluddin.

Prior to 2020, MDES served an average of 50,000 unemployment claimants per year, but it served more than 400,000 unemployment claimants during the COVID-19 pandemic using AWS. In May 2020 alone, MDES served 230,000 people, a historic high and an increase of 4,000 percent from its norm. “We were able to scale up in less than 1 hour,” says Jalaluddin. “On premises, that would have been a nightmare.”

Using AWS, MDES created a core system that meets federal and state-specific security regulations. It uses Amazon CloudWatch—which companies can use to detect anomalous behavior in environments, set alarms, visualize logs and metrics side by side, and more to keep applications running smoothly—to monitor Amazon EKS containers, scaling up and down automatically. MDES also uses CloudEndure Disaster Recovery, which minimizes downtime and data loss by providing fast, reliable recovery of physical, virtual, and cloud-based servers to the cloud on AWS. ReEmployUSA is hosted on a Federal Risk and Authorization Management Program–authorized moderate environment on AWS. “The risk factor is considerably less because all the data is encrypted in the cloud and no personally identifiable information is stored on premises,” says Jalaluddin.

By migrating to AWS, MDES implemented its plan for a shared UI system while optimizing costs and increasing scalability to better serve residents in participating states. The agency expanded the offerings of the system by using robotic process automation and an artificial intelligence–based chatbot. As the agency adds more states to the consortium, it will rely on containerization supported by Amazon EKS. “We guided other states in this project, and we were able to use AWS to meet expanding needs,” says Jalaluddin. “We amended our AWS contract to create a sustainable model as we bring in more workloads and continue the modernization effort. Being able to rely on the same vendor has been invaluable.”