Engineering firm Danfoss has been an HPE GreenLake customer for over five years, and is using its as-a-service model to support its drive to improve energy efficiency in its datacentre environment.
“HPE started with everything as a service, and we have tapped into that,” says Sune Tornbo Baastrup, senior vice-president and chief information officer (CIO) at Danfoss.
The company began planning in 2017, and went live in 2019. Like many industrial firms, Danfoss has a mixed environment. “Today, we’re running Salesforce, and we have some Azure and Google Cloud workloads,” he says. “We also have some private cloud environments and a large legacy IT footprint.”
But migrating everything to the public cloud brings its challenges. “When I asked my CIO peers about their datacentres, we all know it’s very, very difficult to transform them to get into the cloud because there are workloads that have latency dependencies or there is a hard binding to machinery,” says Baastrup, adding that Danfoss wanted to gain the efficiency of the public cloud for the workloads that could not be migrated, which is why the company chose GreenLake.
He believes energy efficiency is now a topic most companies running datacentres are focusing on, and that Danfoss builds some of the most advanced cooling technologies in the world. “We cool almost everything of a certain size and upwards,” he says.
Datacentre cooling requirements are directly linked to the heat produced by the load put on servers, storage and networking hardware. “The way to look at energy should start with reducing consumption,” adds Baastrup. “If we don’t start with reducing consumption, we will not be able to provide enough energy.
“Then we should reuse the energy that we already have in the ecosystem, and we should make sure that the energy we use is green.”
Although datacentres are generally built with capacity to grow, Baastrup believes it’s more important to focus on reducing energy consumption compared with building additional capacity. “It’s not a matter of stopping additional capacity, but we also need to address the energy reduction side, too,” he says.
This starts with ensuring that, at the front end, there is a focus to reduce the energy consumption of cooling equipment, which Baastrup says should begin with optimising workloads. “Make sure you renovate and consolidate your workloads,” he says.
In the Danfoss datacentre environment, Baastrup says HPE has helped the company optimise workloads, enabling the IT team to run fewer servers. The company says IT strategy is aligned with its GreenLake strategy, which is focused on server consolidation.
There is also a server renewal strategy. “We want to get the workloads onto the right tech stacks, which reduce energy consumption and running costs,” he says, adding that there are no trade-offs when tackling energy efficiency in the datacentre. “As we consume less energy, we have less cost and we consume less datacentre capacity, which gives us a little more headroom in our growth journey.”
Typically, when provisioning datacentre capacity, Baastrup says: “We overprovision and, just to be sure, we overprovision significantly.”
However, having an energy savings target means overprovisioning needs to be weighed up against running workloads efficiently. By lowering overprovisioning estimates and running the workloads on GreenLake, he adds: “The analysis HPE can do helps us assess sizing capacity and uptime.”
GreenLake also offers a sustainability dashboard and green practices, which he says helps Danfoss understand the best scenarios for deploying and operating its technology stacks in an energy-efficient way.
