Esper improves performance by 25% and lowers costs by 30% using AWS Graviton

By 2023, Esper was entering a period of rapid expansion. The company was supporting close to a million devices across global customers—point-of-sale terminals, kiosks, walkie-talkies, and IoT endpoints that operate continuously and generate constant API activity. As this footprint grew, so did the performance expectations from enterprise customers who rely on near real-time faster page loads and quicker API responses,” noted Rakesh Chaurasia, senior engineering manager, Platform SRE and DevSecOps at Esper.

Scaling this fast meant Esper needed an infrastructure foundation that could keep pace with unpredictable workloads triggered by customer actions—sometimes large operations executed across thousands of devices simultaneously. The business also needed to maintain high availability around the clock, as devices send updates and events 24/7 across geographies. 

At the same time, Esper’s leadership team was focused on improving operational efficiency to support long-term profitability. With a growing global customer base, the engineering organization needed to optimize resource consumption, reduce operational overhead, and free teams from routine maintenance so they could concentrate on innovation and new capabilities, including AI initiatives. As Raghuveer Chakravarthi, head of engineering at Esper, explained, the goal was to reinvest efficiency gains directly into accelerating product development.

Instead of attempting a full rearchitecture, Esper adopted a bridge strategy on Amazon Web Services (AWS) that combined its existing containerized architecture with a controlled, incremental migration plan. This approach allowed the engineering team to modernize at speed while preserving business continuity and avoiding disruptions to nearly a million always-on devices.

Esper chose to move its containerized workloads from existing x86-based Amazon EC2 instances to AWS Graviton-based EC2 instances to improve price–performance efficiency. “It wasn’t just about cost. AWS Graviton gave us better hardware, more optimized performance, and lower prices compared to the M5 instance family. It was a win-win,” says Raghuveer.

The company uses Amazon Elastic Kubernetes Service (Amazon EKS) to orchestrate microservices and relies on Amazon Relational Database Service (Amazon RDS) and Graviton-based Amazon ElastiCache to maintain availability and reduce operational overhead. For integrated security, Esper uses AWS WAF to filter API requests from millions of connected devices, Amazon GuardDuty to monitor for threats across regions, and AWS Key Management Service (AWS KMS) to encrypt data at rest in alignment with compliance requirements.

Esper launched the full migration in October 2024, moving more than 350 Amazon EC2 instances through a measured, low-risk rollout. “We started small, migrating just two percent of workloads to AWS Graviton while running x86 in parallel,” explains Rakesh. “Once we saw better performance and no issues, we ramped up to five percent, then quickly scaled to 90 percent. The same build ran seamlessly on both architectures, which ensured business continuity.”

By migrating to AWS Graviton-based EC2 instances, Esper reduced compute costs by 30 percent—exceeding its target of 25 percent—and now saves $250,000 annually. The team also lowered CPU core requirements by 20–30 percent. “Every dollar saved is every dollar earned. These savings go straight back into the business—helping us acquire more customers and accelerate growth,” says Rakesh.

Furthermore, the company simultaneously improved application performance by up to 25 percent and reduced API response times by 20 percent, resulting in faster page loads and lower latency across nearly one million devices. “Using AWS Graviton-based EC2 instances, we handle the same workload with fewer servers. Customers get faster responses, and we scale more efficiently,” says Raghuveer.

Looking ahead, Esper is exploring Amazon Bedrock for generative AI use cases and is migrating monitoring workloads to AWS Managed Service for Prometheus and Amazon Managed Grafana to further reduce operational overhead. “AWS helps us spend less and grow faster. The path to profitability is much higher now, and our customers are seeing the benefits every day,” says Raghuveer.