BSS/OSS Cloud Transformation

A national telecom operator serving 12M subscribers across the GCC was operating on a 15-year-old monolithic BSS/OSS stack that had become a strategic liability. Six-month deployment cycles for new services. Three to four major outages per year affecting millions of customers. A billing system so rigid that launching a new product required 4 months of engineering time. And a complete absence of real-time subscriber analytics.

The Challenge

The operator needed to compete with agile digital-native MVNOs and meet evolving regulatory requirements for real-time charging, fraud reporting, and subscriber data management. The transformation required decomposing a 2M-line monolithic codebase into independently deployable microservices, migrating real-time charging without billing errors, and building AI-driven churn prediction — all while maintaining service continuity for 12M active subscribers.

The Strangler Fig Strategy

Diwansoft applied the Strangler Fig pattern — incrementally replacing monolith components with microservices while the existing system handles remaining traffic. This eliminated the need for a high-risk big-bang migration:

Stream 1 — Real-time Charging (Months 1–8). Extracted the rating and charging engine as a cloud-native microservice handling 50,000+ transactions per second. Built on Apache Kafka for event streaming with sub-100ms charging latency — versus the legacy system's 2-second average. Zero billing errors during the cutover, validated across 3 billing cycles.

Stream 2 — Customer 360 & AI (Months 3–12). Built a real-time subscriber analytics platform ingesting call records, data usage, and interaction events. Fed an AI churn prediction model trained on 36 months of subscriber behavior, achieving 87% accuracy in identifying at-risk subscribers 30 days before churn. Proactive retention campaigns triggered by the model reduced churn 23%.

Stream 3 — Product Catalog & Orchestration (Months 8–18). Rebuilt the product catalog as a microservice enabling commercial teams to configure and launch new offers without engineering involvement. Reduced new product time-to-market from 4 months to 3 days.

Infrastructure & DevOps Transformation

Parallel to the application decomposition, Diwansoft implemented a cloud-native infrastructure and DevOps capability:

The operations team shrank from 34 FTEs on call rotation to 8 SREs managing a fully automated platform.

• Kubernetes clusters on AWS with auto-scaling for traffic peaks during major events
• GitOps deployment pipelines (ArgoCD) enabling 40+ deployments per day
• Service mesh (Istio) for traffic management, observability, and mTLS encryption
• FinOps practices reducing cloud spend by an additional 31% through right-sizing and reserved capacity

Results

• 58% OPEX reduction — cloud economics plus automation eliminating manual operations overhead
• 12x faster deployments — from 6-month release cycles to daily deployments via GitOps
• 23% churn reduction — AI prediction models enabling proactive retention interventions
• 87% churn prediction accuracy at the 30-day horizon
• Zero billing errors during 18-month migration affecting 12M subscribers
• 99.95% platform availability — versus 99.7% on the legacy system
• 3 days to launch a new product — versus 4 months on the monolith