Making a legacy cloud-native, the Skyloud testimonials
Discover how Mon Petit Placement, a french startup, migrated their infrastructure to the cloud, and more particularly on Kubernetes and how they made their legacy cloud-native.
Energy management in public and private buildings is an increasingly crucial issue, not only to optimize resources, but also to meet regulatory requirements. In KeepFocus’ native Denmark, as in many EU member countries, smart metering is obligatory for all new buildings, and for all major renovations. Why did this Danish company migrate from on-premise to Scaleway? Let’s find out with Henrik Tudborg, the firm’s Head of Development.
KeepFocus is a platform and service for handling all aspects of building management, as well as a tool for visualizing consumption grouped into different segments, so you can go all the way down to individual meters, or then back up to apartments, buildings and entire customers. You can generate specific or standardized reports, to check things like whether you’re on track with your CO2 consumption.
In Denmark, it’s a legal requirement that new residential buildings have these metering solutions, and that they be added during major renovations (of older buildings). We go one step further thanks to our onsite data collection hardware, which allows you to collect data at a frequency of one sample per meter per hour.
KeepFocus now belongs to Danish company Wise Home. All of our installations are Danish, and therefore so is our revenue. We have around 450 customers today, mostly in private housing, and around 50 in other sectors, like manufacturing (private sector) and municipalities (public sector). In the public sector, municipalities usually need an energy overview of their entire real estate fleet (in this case, non-residential, so offices, kindergartens and so on). In the private sector, we work with both small and large housing customers, who manage anything from 10 to 7000 flats, as well as industrial clients, like factories.
We were on-premise before, with our own data center in the building next to our office. We’d scale up by adding more servers there. When the cloud wave hit, we started experiments with AWS, GCP and Digital Ocean. Then I was hired 4.5 years ago, with the remit to tidy all of this up. We had a lot of microservices, all of them deployed in different ways. We wanted to make sure the deploy story was the same across all our microservices. That resulted in containerizing the entire thing, both for (existing) onsite and the cloud experiments.
It soon became clear that having onsite staff for our own data center was silly, as it wasn’t cost-effective for a small company like ours. So we tried to work out what kind of infrastructure we needed. One of our key requirements was to be able to lease bare metal; doing that with AWS would’ve been very costly, around ten times more expensive than Hetzner or Scaleway.
A lot of our databases had been set up in a very on-premise way: we had two PostgreSQL clusters that mapped onto two physical machines. All of our various databases were all put into the same machine, so different applications were connecting to the same machines. It’s a nice way to operate when you’re a small company; it’s not great if you want to move to AWS, as there you’d just create one managed database per application.
We also had a tonne of data laid out in a way which would’ve required a very big database with the classic cloud providers. Typically, you pick the tier you need, and we needed very fast storage, a decent amount of memory and not that much compute. But with the biggest cloud service providers (CSPs), you can’t pick and choose. So picking the database we needed would’ve been expensive, as we’d also have to have taken a lot of stuff we didn’t need.
Both Hetzner and Scaleway were not only significantly less expensive, but they also allowed us to pick the hardware configuration that suited our needs. The deciding factor for Scaleway was your managed Kubernetes offering; you had a bit more of the managed services we were looking for. Since then, Scaleway has added more and more managed services, some of which we want to try. Furthermore, the way Scaleway develops its products is more ‘cloud-like’ than Hetzner, so it’s more suited to what we need, as that gives us more flexibility. We don’t know what we’ll need in the future, but there’s a greater chance Scaleway will have a managed option for it.
We have 40 different microservice applications that need to talk to each other: that’s a lot of moving parts for a team of five people. The main advantage of a managed service is that Scaleway has to fix it when it goes wrong! That also means we can’t fix it ourselves… But it usually runs well enough for us to be able to orchestrate our applications without having to worry about the machines underneath. If we had just one application, it’d be different! Kubernetes is also useful for the declarative nature of how to orchestrate our applications; being able to abstract things like where an application is, and how to get in touch with it, is super handy for us.
The main difference is the public sector cares a lot more about the details of how and where data is stored. So GDPR is key for them. That was one of the determining factors in our picking a European CSP. It just makes it so much easier to explain to our public sector clients that they can trust that their data is located in Europe, held by a European company …and invisible to the US! If we don’t meet requirements like that, they can’t choose us.
At the time GDPR came into effect, we had a lot of work trying to document all of our sub-processors, and figuring out which of our sub-processors were American and whether we could still use them, or if we needed to find a European provider. We had to audit all of our data and how it was being used by our applications, to make sure the data stayed in Europe at all times.
So now, with personal data, it never crosses a boundary outside of Europe. The only case in which we still use a US provider is for sending data to Amazon S3, once it’s been encrypted on Scaleway servers. This is how we store our database backups, but AWS doesn’t hold the key for those backups. That was a big thing for the public sector. We also had to stop using GCP to proxy data in from machines in our metering fleet (via GCP’s Frankfurt data center), as there was no way to guarantee that data wouldn’t cross into Google in the US.
So today we’ve shut down all activities on AWS & GCP in terms of orchestrating applications. We still have some traffic flowing through US providers, but it’s proxy and encrypted. So effectively we’ve moved all our processing activities to Scaleway.
Reporting is also very important for the public sector. They use our reporting to figure out if they’re on track to reach their goals, for example if their buildings are below a certain threshold in terms of CO2 emissions. We’re capable of translating data from a variety of different sources —including heat pumps, gas heaters, electric heaters or district heating — into one common energy measurement figure.
We’ve also worked on “nudging”, to get tenants to consume less. For example, in one Danish residential block, each apartment has touchscreens which show your energy consumption and how to reduce it, including by comparing yourself with a flat of equivalent size. We’ve found this is way more effective at reducing energy consumption than discovering it in your bill at the end of each month.
Wise Home (which acquired KeepFocus in June 2023) is focused on housing, and billing data for tenants, and this trend is set to continue. This will include integrations with other services, as well as working with associations and smaller housing clients, to help building owners navigate the legal jungle that is currently calculating a bill for a tenant. Heating calculations are particularly challenging here in Denmark (water and electricity less so). So it’s safe to say we’ve got our work cut out!
Discover how Mon Petit Placement, a french startup, migrated their infrastructure to the cloud, and more particularly on Kubernetes and how they made their legacy cloud-native.
Learn how the startup Kanta migrated from OVH to Scaleway, and get the Terraform template to recreate all the bricks of the infrastructure.