Optimising Kubernetes for High Volume Gaming Platform
The Client
Our client operates a large-scale real-time content delivery platform focused on giving the gaming community the chance to express themselves within live streaming ecosystems such as Twitch, Kick, and Discord. The platform enables creators and communities to extend the native emote capabilities of these services through a shared library of more than one million public emotes used across live chat and community interactions.
At its core, our client functions as a high-volume user-generated content and media distribution platform, supporting millions of daily users interacting in highly concurrent, latency-sensitive environments. The service underpins a significant layer of audience engagement within live streaming communities, where emotes form part of the real-time social experience between creators and viewers.
The platform operates on a freemium model, providing open access to core emote functionality while offering premium capabilities such as animated profile avatars, accelerated emote approvals, and subscriber milestone badges to paid users.
From a technical perspective, our client manages substantial global traffic volumes and media delivery demands, processing an estimated two petabytes of data per month across its platform. This requires scalable, performance-oriented infrastructure capable of handling high-throughput content delivery, real-time interaction patterns, and globally distributed user activity.
The Problem
Prior to engaging with DeeperThanBlue, the company had inherited a legacy self-managed Kubernetes environment on bare metal yet didn’t have in-house experience of containerised environments. It was unoptimised, unreliable and vulnerable. The platform was known to fall over at times, sometimes at critical moments when reliability was most needed. Instances such as this had the potential to adversely affect customer experience and damage the company’s reputation. The main challenges faced by the company included:
- Infrastructure Instability: The environment was described as “falling over all the time” and lacked formal support, posing a constant operational risk.
- High Operational Costs: systems were integrated with expensive CDN (Content Delivery Network) partners. Given their two petabyte monthly data transfer usage, these content delivery costs were becoming financially unsustainable.
- Complexity of Self-Management: To avoid expensive potentially pay-as-you-go cloud pricing, the customer had deployed a self-managed version of Kubernetes on bare metal servers, which was difficult to maintain and optimise internally.
What We Built
As a Certified Kubernetes Service Provider (KCSP), DeeperThanBlue was brought in to provide strategic consultancy and technical reconfiguration support. The solution involved:
- Infrastructure Optimisation: We performed a comprehensive service optimisation piece to reconfigure the existing clustered infrastructure for better reliability and performance.
- CDN Migration: To address the cost issues, the team “unpicked” the client from expensive providers and migrated them to Hetzner’s native CDN within the private data centre where the client’s emote library is deployed.
- Managed Support: DeeperThanBlue implemented a formal managed support offering providing the client with a guaranteed Service Level Agreement (SLA) ensuring that Certified Kubernetes Administrators (CKA) are available to respond to issues, typically within two hours.
- Derisking Strategy: The consultancy team unpicked the “container maze” to apply essential security patches and updates, ensuring the underlying infrastructure was no longer a point of failure.
The Results
The intervention transformed our client’s technical environment from a liability into a stable foundation for growth:
- Enhanced Reliability and Availability: The infrastructure is now significantly more stable and available, providing exceptional continuity of service to millions of gamers without the instability previously experienced.
- Significant Cost Savings: Moving away from premium CDN partners to a native data centre solution drastically reduced the monthly expenses associated with their two petabyte data throughput.
- Risk Mitigation: By wrapping the environment in a managed service, the Kubernetes element of the tech stack no longer poses a business risk. Our client now has “insurance” against technical failures through expert monitoring and incident response.
- Operational Peace of Mind: The client’s team no longer needs to manage the complexities of the underlying clustered infrastructure, allowing them to focus on application development and user experience.
- Future Readiness: With a stable foundation now in place, the platform is positioned for further transformation, such as deploying to regions closer to their customer base to further enhance performance and latency.
