WindBorne Systems ensures global fleet reliability with Better Stack

Company: WindBorne Systems

WindBorne Systems is pioneering a new era of weather forecasting. The company designs, builds, and operates a global fleet of autonomous, long-duration weather balloons that navigate the stratosphere. These smart balloons are a global atmospheric sensing system, collecting vast amounts of high-quality data from historically data-sparse regions like oceans.

This novel data source provides a dramatic improvement in the accuracy of weather models. WindBorne's technology, which once surpassed a similar effort by Google, is already being used by leading institutions like the European Centre for Medium-Range Weather Forecasts (ECMWF) and the US Air Force. This has led to tangible results, such as helping produce a 15% more accurate forecast for Hurricane Ian.

Industry: Aerospace / Weather Technology

Use cases: Mission-critical infrastructure monitoring, real-time debugging, incident alerting

Switched from: Fragmented and in-house monitoring tools

Running a global, always-on balloon fleet got too complex to monitor without a single, reliable way to see everything in one place

WindBorne operates a globally distributed, 24/7 autonomous system where operational failure is not an option. Each balloon is a sophisticated, high-value asset, and the data it collects is mission-critical. The company’s cloud infrastructure, responsible for everything from flight control and data ingestion to stakeholder APIs, is the brain of the operation.

As the fleet grew, the team faced a significant challenge: they needed a single, reliable "pane of glass" to observe their entire system. "With a globally distributed, 24/7 running autonomous system, you need to know what’s going on at all times," the team states. Relying on fragmented or manual monitoring methods was not scalable or robust enough to ensure the reliability required for their mission-critical work. The team needed a unified, proactive observability solution.

By integrating Better Stack with Render, WindBorne unified logs and metrics into one product

WindBorne implemented Better Stack for observability, tightly integrated with their Render infrastructure. This combination provides a holistic view of their entire system, from cloud services to the balloons themselves. The solution has two key components:

Unified logs and metrics

To get a complete picture of system health, WindBorne streams all its service logs into Better Stack. With Render’s launch of OpenTelemetry metrics streaming, they are now also integrating critical infrastructure metrics, like CPU and RAM utilization, directly alongside their application logs. This gives the engineering team a single, unified dashboard to correlate performance metrics with system behavior, dramatically simplifying diagnostics.

Real-time incident resolution

Having all this data in one place allows for powerful, real-time debugging. This was proven when the team noticed a critical issue with balloon altitude control.

With unified visibility, the team can quickly spot issues, fix them in-flight, and keep the balloons on track

For a company managing a fleet of high-altitude balloons, knowing what's happening right now is everything. Better Stack delivered the real-time visibility needed to keep the fleet flying safely and collecting critical weather data.

• A unified fleet command view: All flight control logs, satellite commands, and balloon performance metrics now stream into one place, giving engineers a live overview of their entire global operation.
• Flight anomalies diagnosis: When a balloon behaves unexpectedly, the team can immediately investigate the logs to diagnose the root cause, whether it’s a bug in the control algorithm or a faulty sensor reading.
• Critical in-flight intervention: The ability to rapidly find and fix problems, like the gas-venting issue, allows the team to make crucial adjustments mid-flight, protecting the balloon and ensuring the success of its data-gathering mission.

There was a discrepancy in how our balloons were venting gas to control their altitude. After switching from a larger balloon envelope to a smaller one, the control algorithm didn’t adjust correctly. By exploring logs in Better Stack, we tracked down when and where messages were being sent to the balloon over satellite communications. That’s how we discovered the envelope setting wasn’t being applied at the expected time – and were able to fix the issue.

— Kai Marshland | Co-Founder & CPO

Got an article suggestion? Let us know

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.