Vaillant solves scaling IoT challenges with MongoDB Atlas

THE CHALLENGE

New data, new challenges

It was eight years ago that Vaillant first started connecting its devices to the internet. Its plan was to connect the HVAC industry and make the data available through APIs so any customer, partner or engineer could use it.

It didn’t take long for the company to have over 60,000 IoT-connected devices in the field. Just a year later, the number of devices almost doubled. All these were sending more data than Vaillant ever expected.

“We underestimated the adoption rate, the amount of data, and the timeframe,” says Norbert Koch, a software developer at Vaillant. “We had 70,000 devices generating 600,000 transactions per second. Every state change produced data that was sent and stored in our system. And it piled up quickly.”

Adding so much new data, so quickly, meant scaling Vaillant’s existing Azure CosmosDB database became difficult. The team had to constantly keep provisioning more resources to keep it running.

As data volumes grew, response times slowed and the system crashed regularly. With a limited query language, the team struggled to find and filter the relevant data. It put the promise of predictive maintenance and improved customer experience out of reach.

“Within just a few months, we noticed significant decreases in speed and performance and had to manually scale the database to meet specific requirements,” said Koch. “We quickly realized we needed a better database platform that could partner with us to improve scale and unlock the value of connected devices in the field.”

THE SOLUTION

Connected partners

This new database had to be able to handle the unique demands of IoT time-series data and provide a single view of insights across all collected data.

“As we looked at various database solutions, we initially thought that we wanted to manage the ultimate solution,” said Juergen Stauvermann, Senior Java Software Engineer, Vaillant.

Most of the company’s data is hosted on Microsoft Azure so the team needed a solution that was immediately available on the platform. It also required superior scalability, flexibility and performance to be able to handle the streams of IoT data.

Vaillant was already using MongoDB in other projects and had experienced success. It was then that the team decided to migrate to MongoDB Atlas, the global cloud database service.

“We started the process by trying to build our own MongoDB cluster, but our teams are relatively small so we only had a limited amount of time for operational tasks. This influenced our decision to go with a fully-managed solution, available to us as a cloud service.”

Once the decision was made, the team was able to plan and redesign its existing schema. It meant that the data migration took all of 90 minutes.

Using MongoDB Atlas made the difference. “Today, we have over 1.9 billion records and there has been no downtime at all, and our customers immediately noticed the performance improvements,” said Koch. “The whole experience was fantastic.”

“We don’t have to manually scale the solution either,” said Stauvermann. “Everything runs much better on MongoDB Atlas. In the past, we had to manage write limits while scaling the database up or down to keep all systems running. More often than not, we would hit write limits in the evenings, making our systems go down overnight. But not anymore.”

THE RESULTS

Blending the physical and digital worlds

After the switchover, MongoDB Atlas on Microsoft Azure delivers four primary benefits:

1. High-Speed Data Ingest and Real-Time Analytics: Vaillant’s IoT applications can ingest massive volumes of data. With MongoDB’s aggregation pipeline, sensor data is processed to provide the team with real-time analysis, allowing them to build predictive maintenance programmes.
2. Intelligent Edge: Instead of sending all data to the cloud, Vaillant’s edge devices perform some analytics locally which helps to reduce overall network traffic.
3. Fast, Flexible Development Experience: MongoDB flexible data model accommodates complex and quickly changing time-series data as Vaillant adds more sensors to its field devices.
4. Keep Data Secure: With Vaillant’s high-security standards, MongoDB Atlas’ VNet peering-only mode ensures that data is kept secure and access is limited. It can be managed easily so teams still can share and access data when required.

Today, MongoDB’s flexible data model and scalable architecture creates a single view platform of all Vaillant’s disparate IoT data sources - more than 1.9 billion records. It can all be easily accessed, managed and audited by staff or regulators, wherever they are.

With MongoDB Atlas, powerful secondary indexing, and the aggregation framework, the team has decreased its database size by over 99% all the while increasing the throughput of its API by 250 percent. It means Vaillant can provide the kind of IoT services it envisioned all those years ago.

Following the migration, Vaillant launched serviceASSIST, an app that allows installers to remotely troubleshoot, and manage HVAC systems from any iOS or Android smartphone. It saves time, money, and promotes efficiency leading to higher customer satisfaction.

The team now plans to extend its mobile capabilities further with Atlas App Services’s intuitive app development services. It will all help support a new mobile feature, Guided Repair, that will assist remote engineers with real-time data and solutions in the field.

Vaillant also has its sights on analytics using MongoDB Charts, a visualization tool for MongoDB data. It will help map Vaillant’s guidance maintenance, working with AI and machine learning to calculate climate and the weather to make devices even more efficient.

146 years on from its inception, Vaillant remains the market innovator. Now it's backed by the true promise of IoT with plenty more opportunities on the horizon.