Telecommunications
Case Study: Circles Japan Expansion (KDDI povo 2.0)
How shift-left performance validation de-risked a national digital telco launch and supported scaling to 1M+ subscribers in year one
- Japan (APAC expansion from Singapore HQ)
- Microservices
- RabbitMQ
- Cloud-native
- 2020-2022
Subscribers in year one: 1M+
Engagement summary
| Phase | Challenge | Result |
|---|---|---|
| Phase 1 - Pre-Launch Validation | National-scale digital telco launch with untested async architecture | Onboarding, payments, and microservices validated under peak load |
| Phase 2 - RabbitMQ Bottleneck | Prefetch misconfiguration causing uneven worker load under stress | Critical async issue resolved before production go-live |
| Phase 3 - Launch & Rapid Scale | Sustaining performance through 0 to 1M subscriber growth | No major performance incidents during initial adoption phase |
Phase 1 - Pre-Launch Scalability Validation
Circles partnered with KDDI, Japan's largest telecom operator, to launch povo 2.0 — a fully digital, app-driven mobile brand built on the Circles-X SaaS platform. Before the 2021 national launch, the LTE Team supported end-to-end performance validation across customer onboarding, payment and provisioning systems, and a microservices architecture with RabbitMQ-based async workloads.
What we delivered
- Load testing of customer onboarding flows under production-like conditions
- Performance validation of payment and provisioning systems
- Stress testing of microservices architecture and async messaging workloads
- Peak load simulation for national-scale launch traffic
- Observability review and production readiness assessment
Key outcomes
- Stable onboarding performance under national-scale traffic simulations
- Successful validation of microservice and async architecture at peak load
- Confident scaling strategy for marketing-driven traffic spikes
Phase 2 - RabbitMQ Prefetch Bottleneck
During high-scale load testing before the Japan launch, a critical issue surfaced in async processing: RabbitMQ consumers had incorrect prefetch configuration. Workers were reserving too many messages per consumer, causing uneven load distribution — some workers overloaded while others sat idle, driving up queue latency and end-to-end onboarding delays under stress.
What we delivered
- Root-cause analysis of prefetch misconfiguration and worker imbalance
- Tuned RabbitMQ prefetch settings and balanced consumer concurrency
- Re-ran stress and soak tests to confirm stable distribution
- Added queue observability metrics for lag, message age, and throughput
Key outcomes
- Critical async messaging bottleneck identified and resolved before launch
- Stable message distribution under peak load across worker nodes
- Predictable scaling behavior for burst traffic conditions
Phase 3 - Launch and Rapid Scale
povo 2.0 launched in 2021 in collaboration with KDDI Digital Life, positioned as Japan's first fully digital telco brand. The platform faced rapid adoption — scaling from zero to over one million subscribers within the first year, with roughly 70% new-to-operator customers — while maintaining performance stability through the growth phase.
What we delivered
- Continued performance validation through the rapid adoption phase
- Production behavior modeling calibrated for the Japan market
- Observability-driven performance monitoring for ongoing scale
Key outcomes
- Supported launch and scaling to 1M+ subscribers in the first year
- No major performance incidents during the initial growth phase
- Stable platform performance maintained through rapid subscriber adoption
Sectors we have supported
- Entertainment & ticketing
- Media & broadcasting
- Theatre & live events
- Finance
- Telecommunications
- Cosmetics & retail
- Mobile apps
- Healthcare
- Banking
- Oil & gas