Naftemporiki TV
Built a fault-tolerant data pipeline and API system for a leading Greek TV channel, handling real-time stock data, weather info, and commodity prices with zero downtime requirements for live broadcast graphics.
Overview
Zero-Downtime Data Infrastructure for Live TV
Naftemporiki TV, a leading Greek television channel, needed a complete overhaul of their data infrastructure to support their new on-air graphics system. The challenge was building a system that could reliably serve real-time financial data, weather information, and commodity prices during live broadcasts—where even a few seconds of downtime would be visible to thousands of viewers.
I architected and built a fault-tolerant data pipeline that integrates with multiple third-party providers, including the Greek Stock Market's legacy Inbroker APIs, weather services, and commodity data sources. The system needed to handle provider failures gracefully while ensuring data was always available for the graphics overlay system.
Using Hono with Bun for maximum performance and PostgreSQL for reliable data storage, I created an API that independently syncs data from all providers while maintaining 100% uptime. The system includes intelligent fallback mechanisms, stale data detection, and real-time alerting for any provider issues.
Impact
100%
Uptime Guarantee
0
Broadcast Failures
24/7
Live Data Sync
<100ms
API Response Time
Technologies Used
Runtime & Framework
Database
Google Cloud
Integration
Key Feature
Intelligent Fault Tolerance System
Built a sophisticated fault tolerance system that monitors all data providers in real-time. When a provider fails or returns stale data, the system automatically serves the most recent valid data while triggering alerts. This ensures graphics always display accurate information, even during provider outages.
Key Challenges & Solutions
Legacy System Integration
Integrated with the Greek Stock Market's legacy Inbroker APIs, which required reverse-engineering undocumented protocols and building custom parsers. Developed robust error handling to manage the unpredictable nature of the legacy system.
Efficient Data Structures
Designed highly optimized data structures for storing and querying financial market data. The database schema needed to support rapid updates while maintaining fast read times for the real-time graphics system pulling data continuously.
Zero-Downtime Architecture
Architected the system to guarantee 100% uptime for live television broadcasts. Implemented independent sync processes for each data provider, ensuring that one provider's failure never affects the availability of other data sources.
Real-Time Monitoring
Set up comprehensive monitoring with Google Cloud Alerts to immediately notify the team of any provider issues. Implemented health checks, data freshness validation, and automatic alerting for any anomalies in the data pipeline.
Serverless Scalability
Deployed on Google Cloud Run with Cloud Scheduler and Cloud Tasks for automatic scaling and reliable scheduling. The serverless architecture ensures cost-efficiency while handling peak loads during market hours without manual intervention.