PerfectScale by DoiT helps OneFootball optimize Kubernetes for global football traffic at scale
- 25%
- reduction in Kubernetes infrastructure costs
- 80%
- reduction in engineering effort spent on Kubernetes cost optimization and resiliency tuning
After migrating from EC2 to AWS EKS across more than 15 clusters, Rapyd intentionally over-provisioned Kubernetes resources to ensure availability and resilience during the transition. This safety-first approach prevented them from leveraging the elasticity benefits of Kubernetes. Existing observability solutions lacked the granularity needed to right-size workloads, leaving the DevOps team unable to identify the extent of resource wastage across nodes, pods, and clusters.
Rapyd adopted PerfectScale by DoiT to bring AI-driven optimization to their Kubernetes environment. Using PodFit for workload right-sizing and InfraFit for optimal node type selection, the DevOps team integrated PerfectScale early in their service lifecycle—letting new services run for a few days before conducting evaluation and optimization cycles. PerfectScale's pre-configured alerts, severity-based categorization, and Slack integration further enabled proactive issue resolution.
The initial implementation of PerfectScale to our global environments was a revelation. It exposed the extent of our resource wastage and unnecessary expenditure. This realization was eye-opening; we discovered that our provisioning had been excessive across nodes, pods, and clusters, revealing a significant deviation from Kubernetes best practices. We decided to proceed with a Proof of Concept with PerfectScale, embarking on a path to efficient scaling and optimization.
Boris Isakov, DevOps Team Leader and Technical Kubernetes Lead, Rapyd
Rapyd stands out in the fintech landscape by unifying payment technologies and networks into a single platform. They enable businesses to accept, process, and disburse payments globally, supporting over 100 countries for payments acceptance and facilitating payouts in over 190 countries. Their services facilitate sending and receiving funds across the globe, integrating numerous payment methods, and issuing both virtual and physical cards. Additionally, Rapyd simplifies financial operations including money management and foreign exchange, alongside creating virtual accounts to streamline global commerce operations.
Kubernetes plays a critical role in Rapyd's infrastructure, enabling the fintech leader to maintain fast response times which are uniquely crucial in the payment industry. As Rapyd experiences rapid growth, the scalable and dynamic nature of Kubernetes, particularly through AWS EKS across more than 15 clusters, supports their expanding needs. This technology allows Rapyd to swiftly onboard new services and address customer requests efficiently. Rapyd's DevOps group spans three teams—CI/CD, Infrastructure, and Site Reliability Engineering (SRE)—and leverages PerfectScale by DoiT for optimizing their Kubernetes environment, with Boris Isakov, DevOps Team Leader and Technical Kubernetes Lead, overseeing Kubernetes-related activities.
The transition from EC2 to EKS presented Rapyd with a unique set of challenges. The primary goal was to optimize the cloud environment for cost and performance without compromising on the resilience essential for fintech operations. Initially, Rapyd's migration strategy prioritized safety and stability—services moved to Kubernetes were purposely over-provisioned to ensure availability. The team knew this was preventing them from leveraging the full benefits of Kubernetes elasticity. They attempted to utilize observability solutions for environment right-sizing, but soon realized that these tools fell short in providing the granularity needed for effective optimization.
The incorporation of PerfectScale into Rapyd's optimization workflow marked a pivotal shift in their approach to Kubernetes management. PerfectScale's suite of tools, including PodFit for workload right-sizing and InfraFit for selecting optimal node types, empowered Rapyd to refine their resource utilization. By integrating PerfectScale early in their service lifecycle—letting new services run for a few days before conducting an evaluation and optimization cycle—Rapyd is projected to achieve a 35-40% reduction in cloud spending upon completion of their migration and optimization efforts. Beyond the financial gains, the optimization efforts had a profound impact on the performance of Rapyd's services, streamlining operations while improving the performance and reliability of its payment processing platform.
One of the standout features Rapyd leveraged was PerfectScale's advanced alerting system. This capability allows Rapyd to proactively identify and address potential issues within their Kubernetes environment before they can affect service quality. By integrating PerfectScale's alerts into their operational workflows—including critical out-of-memory alerts not surfaced by other tools—Rapyd ensures continuous service reliability and uptime, essential attributes in the fintech sector where even minor disruptions can have significant repercussions for users.
The collaboration between Rapyd and PerfectScale goes beyond a traditional vendor-client relationship; it is a strategic partnership characterized by mutual learning and shared goals. The PerfectScale team provides support and expertise, guiding Rapyd through the intricacies of Kubernetes optimization with a hands-on approach. PerfectScale's deep understanding of Kubernetes best practices and dedication to Rapyd's success play a critical role in the successful implementation of the optimization strategies.
Explore how PerfectScale helps teams right-size clusters, reduce waste, and improve performance without manual tuning.