Data center demand is outstripping supply for the third year in a row. Power is increasingly hard to come by. New cooling solutions and server designs are revolutionizing efficiency. At the same time data centers from the turn of the century are reaching the end of their useful lives. There has never been a better time to upgrade and retrofit, turning old into new.

With Penta Infra having retrofitted nine data centers, I thought some observations on the retrofit process might be of interest to colleagues and customers.

Efficiency where it is needed most

Retrofits transform end-of-life data centers into good-as-new condition, rescuing critical data functions for users. This is what sustainability is all about. They make them fit for the future and far more flexible. As PUE levels improve across Europe, everyone knows that it is the older facilities that struggle to keep up, and they bring overall efficiency levels down. Retrofitting addresses this issue and turns it around.

A second life

Retrofitting also eliminates the huge amounts of embodied impact involved in a new build, reducing the need for new materials, as well as minimizing construction waste streams. Where it’s an existing data center that is being reused there are even more efficiency gains as new equipment takes the compute strain and new users are moved in to improve the ratio of operational power to power overhead.

Asking the right questions

If you are researching an existing facility as a possible retrofit you need a lot of information. Is the current power supply adequate or scalable, or does it need to be replaced, and if so how long might that take? What is renewables availability like locally? Most of the equipment will be end-of-life and you will need to have a detailed replacement strategy and a good knowledge of equipment options, installation procedures, and delivery times. Does the campus have additional development potential? Any redevelopment needs to meet the latest planning regulations, such as height limits or floor space limitations, access, or noise levels. Talking to the municipality and listening to their feedback is key.

Embracing complexity

Although services are already in place, a retrofit is often more complex than a new build. An engineering background is useful as it gives you the tools to plan efficiency optimization at several levels during site selection and development covering everything from power configuration to material and heat reuse and grid integration.

Migration is particularly complex but very necessary. A 20-year-old facility cannot house servers due to today’s rack density levels, and sometimes they contain multiple customers too which makes it more complicated still. Then you really have to work together with the commercial side of the company and start discussing timeframes and mission-critical factors with customers. You can upgrade in a phased way, using new power feeds for existing equipment. The important thing is to develop, test, and agree on a plan together with the operations team, the account management team, and the clients. This makes the facility development team a guest in a live data center, and you need to make sure you are a welcome new member of the team that adds value for customers.

Right-sizing & reuse

The data center industry is very good at committing a lot of capital and therefore, a lot of natural resources to data centers that are barely used, and the level of redundancy is frequently excessive. I don't think that is a sustainable way of thinking. If you use a reliable power grid then your design standards and processes should ensure optimum uptimes without overspecification. If you need cooling equipment, find the most efficient way of designing it in terms of redundancy. Will it be N or N+1 and how will you divide up redundancy? Right-sizing reduces outlay but it's also a sustainable design issue.

Being better neighbors

In a retrofit environment, you are often closer to businesses and residential areas, and you can try to support their needs, which makes you a better neighbor. New power and cooling infrastructure make it possible to transform your site’s relationship with the neighborhood, sharing excess heat or supporting the local grid. If you are installing a heat exchange, you can switch around the mechanics of the heat exchange on the heat pump and it becomes a free cooler. This has the added benefit of making a lot less noise. After two upgrade phases, the second of which increased heat reuse to almost 100 percent, reuse efficiency is now 0.21 in our CPH-1 facility, which means that we are way beyond a 1.1 or 1.2 PUE.

The other option is grid support. Having your diesel generators or batteries as backup for the grid requires code compliance and means you can add resilience to the grid during peaks in demand. You can give something back instead of just installing a data center and building a big fence around it. We offer grid support in Copenhagen and Brussels and are building it out in Hamburg, It’s something we will install in all our new developments, both brown and greenfield.

Phased planning

Long-term plans for brownfield sites are vital. Can you develop the site even further over several phases? How might power needs and providers evolve? Where you have larger sites, as we do in Berlin and Paris, different elements require different approaches. On those sites, there are one or two operational data centers as well as other buildings that can become part of a mixed development and be separated. Or part of the plot can be set aside as a green community space or for biodiversity.

Detaching projects makes them more easy to handle and adds to local amenities. This is where your contractors and partnerships are key to success. In a complex development, you need to run jobs in parallel, so you don’t want to approach it like “half a man and a horse’s head” as the saying goes in Dutch (“anderhalve man en een paardenkop”). Detailed scenario planning is the key, and it needs to take place early and cover the long term.

Reaping the rewards

The rewards of an effective retrofit are significant, particularly in today’s power-constrained and highly regulated hubs. For instance, at our new site in Haarlem, on the edge of Amsterdam, we have a well-used site that has never been fully utilized. By building in the latest sustainable standards we can stay within the current regulatory guidelines and transform efficiency levels.

The end result will be a state-of-the-art data center with a 10MvA connection, allowing 8MW of IT capacity, in a site where demand is extremely high and supply is constrained. Retrofitting is a critical process for both digital business continuity and sustainability.