Containerizing Legacy Applications on Red Hat OpenShift

Executive Summary

A leading East African telecommunications provider operating in 5 countries with 40 million subscribers faced a modernization imperative. Their revenue-generating applications — including the mobile money platform, customer self-service portal, and real-time billing system — were monolithic Java applications running on bare-metal servers. Deployment cycles took 2–4 weeks, and a single bug fix required coordinating deployments across all 5 countries.

The company engaged Cloudix Training's Red Hat practice to containerize 12 core applications onto Red Hat OpenShift Container Platform (OCP). Over 8 months, the team delivered a fully automated CI/CD pipeline, reduced deployment cycles to under 2 hours, and achieved 99.95% availability across all modernized applications.

The Challenge

The telecommunications industry operates on razor-thin margins with massive scale. This operator processed:

• 4.2 million mobile money transactions per day.
• 800,000 concurrent users on the customer self-service portal during peak hours.
• 12 billion CDR (Call Detail Records) per month for billing.
• Real-time fraud detection across 5 countries.

The monolithic architecture created several critical problems:

• Deployment velocity: Any code change required a full application rebuild, QA cycle, and coordinated release across 5 countries. Average time from code commit to production: 3 weeks.
• Scaling limitations: The bare-metal servers were sized for peak capacity, resulting in 70% resource waste during off-peak hours.
• Blast radius: A bug in one module (e.g., the notification service) could crash the entire application, affecting all 40 million subscribers.
• Technology lock-in: The applications were tied to specific JBoss EAP versions running on RHEL 7, preventing adoption of modern Java frameworks and libraries.

Why Red Hat OpenShift?

After evaluating Kubernetes (vanilla), Amazon EKS, and Red Hat OpenShift, the team selected OpenShift for the following reasons:

• Enterprise support: Red Hat's 24/7 support with guaranteed SLAs was critical for a telecommunications company processing real-time financial transactions.
• Integrated CI/CD: OpenShift Pipelines (based on Tekton) and OpenShift GitOps (based on ArgoCD) provided a complete deployment pipeline without needing to stitch together disparate tools.
• Security-first: OpenShift's Security Context Constraints (SCCs) enforce pod security policies by default — containers run as non-root, preventing a class of vulnerabilities common in vanilla Kubernetes.
• Operator ecosystem: OpenShift Operators for Kafka, PostgreSQL, Prometheus, and Elasticsearch simplified the deployment and management of supporting infrastructure.
• Multi-cluster management: Red Hat Advanced Cluster Management (ACM) enabled centralized management of OpenShift clusters across all 5 countries from a single console.

Migration Approach: The Strangler Fig Pattern

Rather than attempting a risky 'big bang' migration, the team adopted the Strangler Fig Pattern — progressively replacing individual modules of the monolith with containerized microservices while the monolith continues to run.

Phase 1: Extract Stateless Services (Months 1–3)

The first candidates for extraction were stateless services with clear API boundaries:

1. Notification Service: SMS, push notifications, and email alerts.
2. Customer Profile API: Read-only service for customer data.
3. Rate Lookup Service: Real-time rate calculation for international calls.
4. Authentication Service: OAuth2/OIDC-based authentication extracted from the monolith.

Phase 2: Extract Stateful Services (Months 4–6)

Stateful services required additional patterns for data management:

1. Mobile Money Ledger: Decomposed into separate credit, debit, and reconciliation services with event sourcing via Apache Kafka.
2. Customer Self-Service Portal: Rebuilt as a React frontend calling containerized backend APIs.
3. Billing Engine: Extracted with a CQRS (Command Query Responsibility Segregation) pattern to separate read and write paths.

Phase 3: Core Systems & Optimization (Months 7–8)

• Migrated real-time fraud detection to a Kafka Streams application running on OpenShift.
• Implemented auto-scaling with Horizontal Pod Autoscaler (HPA) and KEDA for Kafka-based workloads.
• Deployed Istio service mesh for inter-service communication, mTLS, and traffic management.
• Implemented OpenShift GitOps for multi-country deployments — a single Git commit now triggers automatic rollout across all 5 clusters.

Results

• Deployment cycle reduced from 3 weeks to under 2 hours.
• Infrastructure costs reduced by 45% through dynamic auto-scaling (eliminating peak-hour over-provisioning).
• 99.95% availability across all modernized applications (up from 99.2%).
• Mean Time to Recovery (MTTR) reduced from 4 hours to 12 minutes through automated canary rollbacks.
• Developer productivity increased 3x — teams deploy independently without cross-country coordination.
• Zero security incidents in the first 12 months post-migration (compared to 3 in the previous year).

Lessons Learned

1. Start with stateless services. They're the easiest to extract, docker-ize, and deploy — building team confidence before tackling stateful workloads.
2. Invest heavily in observability. Distributed systems are harder to debug than monoliths. The team deployed Prometheus, Grafana, Jaeger, and the EFK stack before migrating the first service.
3. Standardize your base images. The team created golden container images based on Red Hat UBI (Universal Base Image) with pre-configured security settings, logging, and health check endpoints.
4. Train your teams. The development teams completed Cloudix's 5-day Red Hat OpenShift Administration course before the project kicked off. This investment paid for itself within the first month.
5. Don't underestimate the cultural shift. Moving from waterfall releases every 3 weeks to continuous deployment required a complete rethinking of QA processes, change management, and incident response.

About Cloudix Training

Cloudix Training is an authorized Red Hat training partner offering hands-on OpenShift courses, Kubernetes certification preparation, and custom DevOps training programs. Our instructors are Red Hat Certified Architects with real-world deployment experience in telecommunications, financial services, and government sectors.

Visit our Red Hat Training page to see upcoming course dates or contact us for a custom corporate training engagement.