Kiwigrid: Optimizing the green energy experience with an IoT platform powered by Google Cloud

About Kiwigrid

Kiwigrid aims to help businesses succeed in a converging energy and mobility world. Its IoT energy service platform uses cutting-edge technology and services to help its customers open up new markets, diversify their offerings, scale cross-industry business models, and pull ahead of their competition.

Industries: Energy, Chemicals & Utilities
Location: Germany

Since using Google Kubernetes Engine, Kiwigrid has reduced downtimes and eliminated bottlenecks in pre-production, allowing it to accelerate time to market.

Google Cloud results

  • Enables an always-on IoT platform by significantly reducing maintenance downtime and reducing unplanned downtime to near zero
  • Allows developers to build pre-production environments quickly and eliminate bottlenecks with Google Kubernetes Engine
  • Delivers a cost-effective cloud-based solution by allowing Kiwigrid to pay only for what it uses
  • Loads Kubernetes clusters in just three minutes to achieve higher flexibility in deployment

Features go from test to production in hours, not days

While the energy sector has increasingly become decentralized and renewable over the last decade as demand has risen, the customer journey for purchasing, installing, and operating distributed energy systems at a large scale isn’t a seamless experience just yet. Because green energy is decentralized, meaning that it comes from many different sources and locations, the experience can instead be unnecessarily complicated. Energy tech company Kiwigrid aims to change this. Just as important is the company’s mission to address the challenges of integrating renewable energy into generation systems and further decentralizing energy systems across the electricity, heat, and mobility sectors.

“The ideal situation in our view,” says co-founder Dr. Carsten Bether, speaking broadly of the company’s vision, “is that a customer or business pays a flat rate every month, for which they get one or more leased cars, a heating system, a solar panel system on their roof, plus residual electricity.”

That may sound like a future that’s still far off, but Kiwigrid is already working toward its vision of a 100% distributed-energy system. “Electric cars, solar panel systems, and batteries are all purchased at different locations today,” explains Daniel Kühne, Head of Product Strategy and Marketing at Kiwigrid. “Installation, again, is often done by yet another provider. There are many breaks in the customer journey and lots of things happen offline as well. Our platform is designed to make the experience seamless by bringing together all these elements.”

The German company connects and manages all of the different devices installed in a home or an apartment building, returning them to a centralized state and enabling easier management in one place: the KiwiOS platform. “By using the platform, one of our clients was able to offer a more transparent photovoltaic and battery storage system to its customers,” shares Daniel. “The client can monitor its systems remotely and offer enhanced error management to its customers.” Kiwigrid’s platform also helps to optimize the use of renewable energy, enabling intelligent distribution from multiple energy suppliers.

For Kiwigrid to fulfill its mission of decentralizing the energy world, it needs reliable underlying technology. That’s why Kiwigrid runs its platform on Google Cloud and Google Kubernetes Engine.

“Back in 2017, Google Cloud was the first vendor to offer a managed Kubernetes solution. With other providers, we would have had to manage Kubernetes master and worker nodes manually. For us, there was simply no comparable offer.”

Axel Köhler, Cloud Architect, Kiwigrid

Moving to the cloud for a more scalable approach to infrastructure

Before making the decision to switch to a cloud provider, Kiwigrid was solely hosted on physical servers. This all changed in 2017, when the company received an opportunity to offer hosting in Australia and the question arose of how to scale to international markets.

Axel Köhler, Cloud Architect at Kiwigrid, recalls what he was thinking at the time: “It didn’t seem like a sustainable solution to rent servers in Australia and build the infrastructure from scratch. We needed to find a new approach.”

At the beginning of 2017, the team started to compare its options, from leading cloud providers to self hosting, looking at each one in terms of managed services, availability, and cost-effectiveness.

After the evaluation, Kiwigrid chose Google Cloud as the best option due to its technical strengths, fully managed Google Kubernetes Engine solution, and cost-effectiveness. “Back in 2017, Google Cloud was the first vendor to offer a managed Kubernetes solution,” explains Axel. “With other providers, we would have had to manage Kubernetes master and worker nodes manually. For us, there was simply no comparable offer.”

There was another important reason for collaborating with Google Cloud instead of any other cloud provider. Together, Google Cloud and Kiwigrid make it possible for customers, mostly utilities, to store their data in a hybrid cloud. Google Cloud components are used to process and analyze data anonymously, while a private Kiwigrid data engine is used to securely store them and bring them into context. As Kiwigrid is certified for high-level security in the upcoming smart grid, required for highly secure distributed energy use cases, the use of Google Cloud can also be provided for highly secure, data-sensible regulated use cases.

“The node autoscaling feature in Google Kubernetes Engine is simply unbeatable. It allows us to scale very quickly and to put experiments into operation way faster. It also helps that new clusters take only three minutes to load.”

Axel Köhler, Cloud Architect, Kiwigrid

Getting to market faster, with Google Kubernetes Engine autoscaling

Since its migration in September 2017, Kiwigrid’s development team has gained plenty of flexibility, especially in terms of deployment capabilities. “The node autoscaling feature in Google Kubernetes Engine is simply unbeatable,” says Axel. “It allows us to scale very quickly and put experiments into operation way faster. It also helps that new clusters take only three minutes to load.” In practical terms, this means that Kiwigrid has reduced its lead time by being able to spin up as many pre-production environments as needed, compared to only two before.

Previously, having only two pre-production environments led to many bottlenecks. Eliminating those has allowed Kiwigrid to release new products into the market a lot faster. Axel compares: “It used to take us five to seven days to take our main product from the test systems into production. Today, with Google Kubernetes Engine and continuous delivery pipelines, we’re able to deliver within hours.”

Kiwigrid has also taken advantage of Google Kubernetes Engine to improve the stability of its platform. “We’ve gained higher stability when it comes to both planned and unplanned downtimes, which enables us to improve availability for our platform,” says Axel. “We used to have large periods of downtime due to maintenance, but we’ve been able to reduce those significantly. After redesigning our systems on Google Kubernetes Engine, unplanned downtimes are rare.”

Another positive point for Kiwigrid is that the on-demand cluster creation offered by Google Kubernetes Engine helps it to optimize the cost of Google Kubernetes Engine. Kiwigrid doesn’t have to leave machines running all the time if it only needs them for a short while. This allows the company to be quick and flexible with many pre-production environments, while at the same time only paying for what it uses.

From a developer’s perspective, Axel adds, Google Cloud offers an intuitive experience. “From the moment you open the cloud console, you can start being productive right away,” he says. “I can do what I need to do, easily and quickly. It’s the same with Google Kubernetes Engine, there’s nothing getting in the way of development.”

“We’re confident that once we’ve completed our data management project, we’ll be able to use more Google Cloud products to gain fresh insights. Those findings could help us better analyze the increasingly connected energy sector and optimize the unique interactions of buildings, e-cars, smart grids, and smart meters.”

Axel Köhler, Cloud Architect, Kiwigrid

Using intelligent insights to optimize the grid, with Google BigQuery

Kiwigrid collects data on activity such as outgoing power or voltage at each grid phase, or state of charge for batteries, and this information helps it optimize energy for around 10,000 home energy management systems and 19,000 building energy management systems on its platform. Because each system is normally connected to an average of 10 devices, such as power meters, photovoltaic plants, or heat pumps, Kiwigrid needs a data management and analysis solution that can help it gain valuable insights from tens of thousands of data points.

The company is looking to create a more ML-driven system design for data management, using Google Cloud products such as BigQuery. By storing its data securely in Google BigQuery, Kiwigrid plans to run fast queries and visualize results through products such as Datalab. It’s also experimenting with Dataflow to build an ML pipeline using AI Platform to run Tensorflow ML models that will help to increase performance.

“We want to separate the data that needs to be highly available all the time, from data that we store for disaster recovery or analysis,” says Axel. “This will open new possibilities for analyzing data to gain new insights that will help us make the best decisions for Kiwigrid and our customers. We think the system will also help us uncover opportunities for new business models.”

The company is also exploring IoT Core for device management and distribution of ML models so it can maximize data insights. “Our platform forecasts and measures energy production and consumption, and we’re confident that once we’ve completed our data management project, we’ll be able to use more Google Cloud products to gain fresh insights,” says Axel. “Those findings could help us better analyze the increasingly connected energy sector and optimize the unique interactions of buildings, e-cars, smart grids, and smart meters.”

For Daniel, it always comes back to making the energy-buying journey more seamless for customers. When it comes to the optimization of this new decentralized green energy world, he sees plenty of opportunity for improvement in the future. “In the end, it’s all about the value we’re able to give people who want green energy,” he says. “It’s not an easy transition to make alone, so our job is to step in and bring all the actors together for everyone’s benefit.”

About Kiwigrid

Kiwigrid aims to help businesses succeed in a converging energy and mobility world. Its IoT energy service platform uses cutting-edge technology and services to help its customers open up new markets, diversify their offerings, scale cross-industry business models, and pull ahead of their competition.

Industries: Energy, Chemicals & Utilities
Location: Germany