Paytm Realizes 35% Savings and Up to 30% Application Performance Improvement using AWS Graviton

One97 Communications Limited (OCL) that owns the brand Paytm is India’s leading payments and financial services distribution company and the pioneer of QR, Soundbox, and mobile payments. In its mission to bring 500 million unserved and underserved Indians to the mainstream economy, Paytm runs multiple businesses across payments, loan distribution, wealth management, insurance distribution, and online travel and ticketing etc.

AWS Graviton-based Amazon EC2 instances in AWS Mumbai region (ap-south-1) are priced almost 50 percent lower than comparable instances. Paytm migrated to AWS Graviton to optimize EC2 costs. “When AWS announced the AWS Graviton processors in the Mumbai Region, we were one of the first companies to do a proof of concept,” says Paytm. On Graviton based instances, Paytm saw better price and performance for their applications, while executing a proof of concept and saw the advantages of using AWS Graviton in multiple applications throughout the company.

Paytm first migrated its MySQL workloads in mid-2021, including a self-managed MySQL cluster with over 160 nodes in production, to AWS Graviton-based instances. To do this, Paytm had to first update its MySQL database to a minimum software version recommended for AWS Graviton processors, and because of the criticality of the databases as well as dependencies between workloads, this update was a complex undertaking. “The AWS teams supported us very well, especially considering the scale and speed at which Paytm works,” says Parakh Aggarwal, Director of Cloud Excellence and FinOps. Paytm not only realized a 50% reduction in prices for EC2 instances, but also saw 20–30 percent performance improvements in its MySQL workloads (24% in Transaction Per Second and 30% in Queries Per Second) compared with its previous solution. This workload migration helped Paytm realize the potential savings from Graviton adoption, which laid the foundation for a systematic and program driven expansion of Graviton usage across multiple workloads in Paytm.

For its payment gateway migration, Paytm migrated 80 percent of its more than 1,000 Amazon EC2 instances to AWS Graviton-based instances by December 2023. Nearly 70 microservices were successfully migrated for the payment gateway alone. After migrating this portion of the payment gateway infrastructure, Paytm achieved an overall cost reduction of up to 35% of their Amazon EC2 spends. Additionally, Paytm improved the performance of its payment gateway services for both end users and merchants after migration. The payment gateway’s throughput improved by 47% in over a 6 months period from Mar to Sep 2023, when the Graviton adoption for this workload increased from 8% to 30%. Due to better performance of the workload running on Graviton, Paytm operated the workload more efficiently, leading to an estimated decrease in workload carbon intensity (workload carbon emissions per transaction) up to 47%.

For its data lake, Paytm had been using Amazon EMR—a cloud native big data solution for petabyte-scale data processing, interactive analytics, and machine learning—for around 65 percent of its analytics workloads. Paytm began the migration of these workloads to AWS Graviton-based instances in 2022 and achieved 30–35 percent cost savings. By the end of 2023, 80 percent of Paytm’s Amazon EMR workloads ran on AWS Graviton, resulting in better performance. “One benefit of migrating AWS managed workloads to AWS Graviton-based instances is that, the functional compatibility and performance of these workloads is managed by AWS and, as a result, we have fewer complexities to manage,” says Aggarwal.