Building an Automated Workflow to Process W-2s Using AWS Serverless Solutions with ADP

For 20 years, ADP has been running its Wage Payments business on a mainframe computer, which sometimes glitched when the load of W-2s stressed the system. Additionally, clients were unable to search individual W-2s without downloading and laboriously navigating large PDFs. “Our mainframe system had limitations, which were hurting our business,” says Lohit Sarma, senior vice president of product development at ADP. “It was affecting our development speed and our ability to optimize the user experience. Our speed and service cost were the two primary factors driving us to move away from the legacy stack.”

ADP was already running its human capital management technology using about 250 microservices from AWS. After evaluating potential solutions, ADP elected to rebuild and modernize its Wage Payments system using AWS Lambda. (See figure 1: ADP Wage Payments system on AWS.) In 8 months, the team gained expertise in AWS serverless solutions, rebuilt its system, and migrated the first 100 clients. “When we started this journey, the Wage Payments team was new to the cloud,” says Narendra Verma, director of development for Wage Payments at ADP. “With the help of the AWS team, we upskilled our team. We got up and running on the technology at a speed beyond our expectations.”

Built on AWS Lambda, the modernized system automates the workflow of ingesting, processing, producing forms and reports, and billing. The business uploads its W-2 data to ADP, which automatically extracts information, validates and saves data, and generates forms and reports in a manageable and searchable format. For clients with more than 400,000 employees, the forms and reports download service runs on AWS Fargate, which provides serverless compute for containers. “The report format is a major benefit now,” says Verma. “Earlier, all reports used to be in text format. Now, the readability of reports has improved a lot, which enhances our client experience.”

For document and report storage, ADP uses Amazon Simple Storage Service (Amazon S3), an object storage service offering industry-leading scalability, data availability, security, and performance. Rather than provisioning its database by hand, ADP uses the serverless Amazon Aurora, a relational database management system built for the cloud with full MySQL and PostgreSQL compatibility.

Instead of a monolithic model, ADP uses the agility of AWS Lambda to quickly test and deploy automations in independent environments. “We improved our speed to market,” Verma says. “Our development is really efficient, and we can now do more releases than before.” ADP also regularly betters the 4-hour requirement of its service-level agreements for file processing, and it improved its API response time for clients by 55 percent. “Using AWS, we have seen a 14 times performance improvement in our processing layer at one-third the cost of our mainframe architecture,” says Verma.

In addition to cost savings on processing, ADP no longer has to spend on the maintenance of its legacy mainframe. “It’s not an apples-to-apples comparison because we don’t have to worry about ordering hardware, selecting security patches, and meeting compliance requirements, which all go through research-and-development cycles,” says Sarma. “Now, we can just concentrate on product development while the rest is handled by AWS.”

ADP operates across multiple AWS Regions to maintain virtually 100 percent uptime and robust disaster recovery that is critical for clients, especially during January, the peak month. “The system is designed to be up and running during this critical time, which then helps us sell to our clients,” says Sarma. “They know it’s a reliable system. It’s super, super critical for our system not to go down.”