Romande Energie Reduces Platform Costs by 40% and Makes Substantial Wholesale Energy Savings with Snowflake

Romande Energie produces and supplies energy to thousands of business and private customers across French-speaking Switzerland. In addition to energy supply, the company also sells sustainable energy-producing solutions, such as solar panels and wind turbines, forming part of the company’s overall drive to make its geographical area the first low-carbon region in Switzerland.

Buying and selling energy is a complex business — even small errors can cost millions of dollars. Not only are traders under pressure to buy energy more than a year in advance at the best prices, but they also have to balance supply with real-time consumer demand. This means that even the best-considered decisions can result in undersupply or oversupply, leading to cost inefficiencies and, in the worst case, blackouts. 

To get these decisions right, data is a critical part of any energy supplier’s operations — including Romande Energie, which buys 70% of its supplied electricity. Data-generating smart meters have changed the game for Romande Energie, allowing the company to accurately analyze customer consumption and predict future demand. 

Data also presents a unique opportunity for Romande Energie to improve network efficiency, optimizing everything from infrastructure renewal plans to changing customer consumption behavior. But combining, processing and analyzing countless data sources at scale to produce actionable insights is no easy feat. 

As Nils Rinaldi, Data Team Lead at Romande Energie, explains, having the right data infrastructure in place makes all the difference: “We used to use Cloudera on-premises with one development and one production environment. But the environments were weighted towards production, so when our developers tried to run SQL queries on the development environment, they would often fail. We also couldn’t integrate our platform with SNP Glue, a change data capture software solution, meaning we couldn’t extract data from large tables.” Managing an on-premises platform also created significant maintenance challenges for Romande Energie, requiring its team to perform regular out-of-hours upgrades and fend off increasingly sophisticated cyber attacks. 

What’s more, Romande Energie’s traders rely on up-to-date data that’s available no later than 10 a.m. each day to make accurate energy purchasing decisions. But due to performance limitations and data volumes, they had to make do with day-old data — causing substantial cost implications. 

To ingest, combine, process and analyze wide-ranging data sets at scale, Romande Energie needed a scalable data platform that could offer near-limitless performance without costing the earth.

After evaluating several data platforms in line with the company’s recent cloud-first approach, Romande Energie chose Snowflake’s Data Cloud.

Following a gradual migration to the Data Cloud, including integration with the company’s Microsoft Azure instance, Romande Energie used Snowflake Professional Services to upskill the team on the platform and gain on-demand technical support. “Snowflake Professional Services gave us answers to specific architecture questions,” says Rinaldi. “The Snowflake team was experienced enough to have already dealt with these issues and would come to us with multiple solutions. The service meant we could get up and running far quicker without the need to test and learn different solutions. And it removed the risk of making the wrong decisions.”

Romande Energie now benefits from Zero-Copy Cloning, a tool in Snowflake’s Data Cloud that allows the company to copy a database (typically production database to development database), table or schema without consuming additional storage space or time — and all without duplicating existing data. 

“Sometimes we just want to test new iterative features on the fly,” Rinaldi says. “Without Zero-Copy Cloning, we would risk introducing unnecessary bugs and taking too much time. It’s enhanced our ability to experiment.”

Using Talend for its extract, transform and load (ETL) processes, Romande Energie ingests its smaller table data into Snowflake each day. For larger table ingestion, it uses Snowflake’s seamless integration with SNP Glue. The company first models data, then passes it to Qlik Sense for business intelligence analysis. 

Romande Energie uses Snowflake’s integrations with the company’s other data platforms to help ensure its data is secure and only available to those who really need it. Rinaldi and his team now use Microsoft Entra ID and Azure Key Vault with Snowflake to ensure business-critical data is protected. “Integrating Snowflake with tools in Azure was very easy,” Rinaldi says. “Thanks to the features in Snowflake’s Business Critical Edition, we can build private encrypted links between the two platforms for complete peace of mind. Our data environment would be a no-go without it.” 

For governance, the company uses object tagging in Snowflake to set data access levels and track lineage. Romande Energie even integrates third-party data sets to augment its internal data with external insight. For example, the company has connected multi-location weather data feeds from MétéoSuisse, helping it make better energy consumption and production decisions based on predicted customer demand. 

In platform costs alone, Romande Energie has saved 40% compared to its previous solution. It’s also rebalanced its spread between development and production environments to deliver shorter development cycles. The company's development team can also quickly create new environments whenever they need them to support new machine learning models.

The changes have also meant a more productive smart meter rollout across Romande Energie’s customer network — which ultimately benefits customers’ bills. With all its smart meter data now being fed into the Data Cloud and connected to business intelligence tools, the company has access to critical performance insights across these connected devices. This means that it can prioritize maintenance and engineer callouts for devices that aren’t responding as they should and reduce any disruption to customer billing.

Building on their success so far, Rinaldi and his team are using what they’ve achieved to prove a business case for Snowflake across other parts of the company — particularly around electricity grid infrastructure planning and consumption and production forecasting. They also intend to experiment with and implement improvements to their existing environment. This will include adding self-service business intelligence for easier data access and new data governance features, including developing an operational data hub with Hybrid Tables — Snowflake’s lightweight transactional use case tool. 

A natural next step for Romande Energie is to experiment more with AI. While the company is already using AI in day-to-day operations, it’s keen to take advantage of what Snowflake has to offer. For example, Rinaldi will look at transforming manual document search processes through a large language model, reducing the amount of time people spend searching for and processing documentation across the business.