Data Fellows

Production systems

A selection of projects where stability, performance and predictability were the main goals. Solutions built for real workloads — not demo environments.

  • High-traffic platforms
  • Legacy systems under control
  • Production-grade database & backend optimization

Sector: Digital media · Realtime content · High concurrency

High-traffic platform

Backend and database optimization for an international digital platform facing extreme peak load during live events.

Situation

A growth platform under peak load

The platform processed large volumes of real-time traffic. Load did not grow linearly — it spiked sharply in short bursts. During live moments, the number of requests quadrupled within minutes.

Symptoms:

  • Timeouts during traffic peaks
  • Unpredictable response times
  • Database waits and lock contention
  • Scaling only provided temporary relief
Approach

No rewrite — targeted interventions

Instead of a full rebuild, we analysed the existing system. The focus was on bottlenecks, query behaviour and request flows through the backend.

  • End-to-end performance analysis (p95/p99 latency)
  • Database query and index optimisation
  • Cache strategy aligned to traffic patterns
  • Restructuring queue and API flow
  • Observability focused on action rather than dashboards
Outcome

Stability under load

The platform remained stable during peak moments where incidents used to occur. Response times became predictable and operational pressure on the team dropped significantly.

  • Significantly lower p95/p99 latency
  • Fewer production incidents
  • No need for overprovisioning
  • A backend that is easier to operate and evolve

Sector: Manufacturing · FMCG · Business-critical databases

Long-term database operations

Continuous monitoring and operations for production databases at an international manufacturing organisation.

Situation

Business-critical systems without a DBA team

The organisation ran multiple operational databases directly connected to production, logistics and internal applications.

Stability mattered more than innovation. Incidents were simply not acceptable, but there was no dedicated DBA team.

Approach

Calm operations instead of project work

Data Fellows took over day-to-day database operations. Monitoring, deployments and troubleshooting ran continuously in the background.

  • 24/7 database monitoring
  • Proactive troubleshooting
  • Release and deployment support
  • Consultancy for new development initiatives
  • Performance monitoring without disrupting production
Outcome

Years of stability

With continuous monitoring and early signal detection, interventions were rarely needed. Systems remained stable while development teams focused on new functionality.

  • Minimal production incidents
  • Predictable operational costs
  • No internal DBA capacity required
  • Fast support for new development

Sector: Digital media & sport · Customer identity · Platform integration

Fan platform integration

Integration of multiple digital platforms into one central customer environment for marketing, ticketing and fan engagement.

Situation

Disconnected platforms

Video, ticketing, webshop and fan environments operated as separate systems. Customer data was spread across multiple applications, which prevented a complete customer view. Customers also had different identities across the individual systems.

Marketing, service and commercial teams worked with partial information as a result.

Approach

Backend integration on Azure

Data Fellows implemented a central identity layer using Microsoft Entra External ID (formerly Azure AD B2C). Fan platforms were connected through one shared account directory. Authentication, profile management and customer data became available across platforms.

  • Integration of video, webshop and ticketing platforms
  • Event-driven integration via Azure Service Bus
  • Customer data normalisation
  • Sync to Dynamics 365
  • Backend optimisation and controlled data flows
Outcome

Customer 360

All interactions were consolidated into one central customer view. Teams gained near real-time insight into fan behaviour, purchases and engagement across channels.

  • One consistent customer identity
  • Better marketing segmentation
  • Less manual synchronisation
  • A stable integration architecture for future expansion

Want to discuss your production systems?

We’ll review risks, validate bottlenecks and provide clear, actionable advice.