How software efficiency delivered a 98.5% cost reduction
Contribution by Maaike Stoops – Head of R&D at LINKIT
Short summary
The cloud offers speed and flexibility. But in practice, we often see costs rising faster than value. That was exactly the case for a high-tech manufacturer in the industrial sector. Within its data platform, one of the core data pipelines was running at approximately €470 per day. After a focused software optimization trajectory, that same pipeline ran at just €7 per day. That translates into an annual saving of over €170,000. The breakthrough did not come from cheaper cloud resources, but from making existing systems work smarter.
Cloud technology creates enormous opportunities for manufacturing companies: instant scalability, experimentation without heavy upfront investments, and faster response to market demand. However, these advantages come with a risk. When data pipelines are built in the cloud without strong architectural governance and software optimisation, costs can spiral out of control quickly.
That is exactly what happened here. What started as a logical move toward flexibility and scalability turned into a situation where costs increased faster than business value. In high-tech manufacturing, where data volumes are massive and margins are tight, control over cloud consumption is critical.
The challenge: rising cloud costs and limited visibility
Every organization wants to become more data-driven. In this case, the company had access to sensor data from hundreds of machines, generating billions of records per month.
Migrating to the cloud initially made sense: rapid scaling, elastic compute power, and innovation potential. But the migration was executed without a thorough code and architecture review.
The consequences:
- Structural overprovisioning of resources
- Overly generous auto-scaling configurations
- Inefficient code consumes excessive CPU and memory
- No operational cost visibility per workload
Cloud invoices kept increasing, while the business saw no proportional value in return. “There was no clear understanding of data behaviour or cost per operation.” explains Wenderson Ferreira de Souza, Advanced Data Architect at LINKIT.
“Infrastructure scaled faster than necessary, and that was immediately reflected in cloud costs.”
The approach: software optimization for maximum efficiency
Our team returned the pipeline to its fundamentals. Measurement was key. We performed a deep analysis on:
- CPU consumption
- Memory utilization
- I/O behavior
- Processing logic
This revealed clear inefficiencies: suboptimal code, oversized configurations, and poorly tuned compute environments.
From there, we implemented a structured optimisation plan:
- Rewriting critical code segments
- Tuning JVM configurations (including memory management)
- Optimising auto-scaling rules to align with real demand
- Iterative performance testing and cost validation
Each improvement cycle was followed by performance and cost measurement. No assumptions, only validated results. To ensure structural improvements, we ran realistic benchmarks under production-like loads. The savings proved sustainable, not temporary.
“By thoroughly analysing the application, we identified exactly where optimisation would deliver impact.” says Wenderson.
“We reconfigured the system to use capacity intelligently. It now runs more stable and more efficient, even under pressure.”
The result: sustainable cost reduction and full cloud cost control
The impact was immediate and measurable. The pipeline that previously cost €470 per day now runs at €7 per day. That equals a 98.5% reduction in daily cloud spend and over €170,000 annual savings, without compromising performance.
“Within weeks, the difference was visible: 98.5% lower daily cloud costs, improved performance, and significantly greater predictability.”
But the impact went beyond financial savings. The organisation developed a new awareness: cloud consumption is not unlimited and should never be unmanaged. Engineering teams began taking ownership of cost behaviour. Software optimisation became more than a technical exercise, it evolved into a cultural shift toward efficiency and sustainability.
By deploying cloud infrastructure more intelligently, servers perform better, and pipelines run faster and more stably.
And that principle extends beyond cloud environments. Wherever systems operate, efficiency at the software layer drives structural performance gains.
