One year after Hydro66 became the first OCP Ready™certified data center in the nordics, OCP VP Channel Development Steve Helvie and Hydro66 trailblazers, Andy Long and Paul Morrison participated in an informal discussion. Topics covered included how Hydro66 embodies the OCP tenets of openness, impact, scale and efficiency and why there is still room for more data center innovation.
Steve, Andy and Paul discuss the climate crisis and COVID-19 as acccelerants for OCP and how Hydro66 are contributing to progress by, for example, operating highly efficient colocation infrastructure using the lowest cost locally generated renewable energy in the EU.
OCP Hydro66 Fireside Chat October 2020 – Chat Transcript
STEVE:
It’s my pleasure today to speak with Andy Long and Paul Morrison from Hydro66. Hydro66 is our very first OCP-ready facility in the Nordics. Paul is the Chief Commercial Officer and Andy is a strategic development manager for Hydro66. So thank you both for taking the time today.
PAUL:
Thank you, Steve. Pleasure to be here.
ANDY:
Good to be here, Steve.
How did you get into the data center industry?
STEVE:
Can you please tell us a little bit about yourself and then how you kind of got into the data center industry, as well as an overview of Hydro66?
ANDY:
Sure. Thanks, Steve. So, I mean, Hydro is a team of industry veterans if you like, my background is technical. You know, I started with a computing degree back in the mists of time and myself and Paul, and the founder of the business all worked together at an internet provider called Easynet through the 90s and into the noughties. And we’ve always been about disrupting industries and trying to improve the way of doing things. So in 2014 we saw an opportunity to disrupt the data center industry. We saw inefficiencies in the marketplace and we wanted to take some of the economies that we see in hyperscaler data center designs and locations and make them available to the broader market.
PAUL:
Yeah, I have a similar background to Andy, but possibly slightly more commercial on the sales and marketing side, but extensive experience in the telecoms, ISP and data center space. So it’s been a very interesting couple of decades I guess seeing how data centers have morphed from on-premise server rooms effectively up to the extremely large hyperscale 21st century factories, like we see today. So it’s been great to be part of that journey being involved in the building of large scale facilities in various parts of the world. And then we have the climate crisis to cope with and perhaps the pandemic as well as an interesting accelerator for our business too. So it feels like there’s always something to change, develop and grow with inside this industry.
Why did you decide to build the data center in the north of Sweden?
STEVE:
Yeah, it’s been interesting to watch the growth specifically for your team. Now, you decided to build in Sweden, Luleå, which is in the north of Sweden. Why did you decide on that particular location? And what makes that area special?
PAUL:
Yeah, it’s a great point and as Andy alluded to earlier, we always have had a slightly disruptive part of our DNA been looking at how we can add value and change things up in the sectors that we’ve been involved in. And back in 2013, early 2014, we were looking at data center investment opportunities and we became aware of what Facebook had done in Luleå to open a hyperscale facility to serve Europe. That was their first data center build of any size outside of North America and from conversations with those guys, and with some of the site selection teams involved, it became clear that they’d conducted a global exercise to look at where in the world was the next best place to build outside of America to serve their audience. And it was fascinating for us to see that they had chosen northern Sweden. So we looked at that, and thought perhaps there’s some ground here for us to do something similar for everyone else. Facebook have built for their own application serving purposes and we thought perhaps there’s something we can do for the general colocation market.
And then narrowing that down, inside northern Sweden, we became aware of a facility that was very close to a brand new substation that was served by a 14-terrawatt hour river system, 4500 Megawatts. And for us, that’s kind of like a data center heaven when you can establish a facility on a piece of ground that’s got access to such large quantities of renewable power. And then you add in the highly skilled workforce in the region, the cool average temperature from a climate perspective and the respect for the environment that that country has, becomes very compelling. And then looking at the cost of power – the overall power situation in Europe general is highly constrained – most national grids are really stressed with the move towards renewables.
So we felt the timing was right to do something quite different and quite radical for the data center space, and that was the ideal place for us to build something different.
ANDY:
Yeah, just expanding on that, one of the things that really stands out for me in the migration away from metro centers is that the drivers that that caused what I would think of as legacy data centers and hosting sites to locate in the big metro centers, at the end of the 90s, in the start, and over the last 10, 15 years were really driven by telco. You know, the reason that in Europe is everything is focussed around Frankfurt, London, Amsterdam, and Paris, and in the US, on the West Coast and East Coast was the high cost of connectivity and so everybody was driven to build hosting facilities next to the IXPs, on the most expensive real estate with the most unreliable power. So it’s not where you would choose to go to with a blank sheet of paper. And what’s happened in the last 5, 10 years is as the cost of telco has fallen exponentially. It’s been much more sensible to take the data to the power, instead of having to take your data into these kind of expensive metro centers. So that’s what’s enabled this big shift. The older industry players are stuck with their long-term capital investments in these really expensive locations where they have to bolt on lots of extra resiliency. But when fiber is so cheap, and the optics are so good, it just makes sense to go to places where the power is 100% renewable, where the temperature is cold, and where there’s great people and great skills.
STEVE:
Yeah, and one of the things I like best is the way that you took a look at what the hyperscalers are doing and then localize that for both the region and the market segments that you wanted to address. So mimicking what Facebook was doing, tailored for a particular market is pretty interesting. So as you look at the market in general, some people may believe that the data center market is a bit overcrowded, so what’s your view on this?
PAUL:
I think there is space for new entrants. Innovation is a real driver for human progress basically. We need innovation and data centers are no different. So we need to innovate what we’re doing. If we really believe that data centers are the 21st century factories, then how many factories do we see in city centers? Not many, right, and the reason for that is, city centers belong to people these days. So we need to get data centers out of city centers and that’s one of our fundamental beliefs and to build this at scale requires that.
Other areas of innovation can be in terms of cooling technologies to get that PUE down as low as possible. Perhaps there’s things that we can do with automation to make maintaining the servers quicker and easier. So there’s plenty to go at in terms of the data center space. If you want to build a me-too data center in a large urban conurbation then perhaps it is overcrowded metaphorically and literally. But we feel that for new market entrants that are not so worn down by the legacy of how things have always been done, then there’s room for new market entrants.
How have customer requirements changed over the last 6 months?
STEVE:
So as you’ve looked at the last six to 12 months, how have customer requirements changed during this quite significant impact economically? As you’ve dealt with customers in the past, how they change their current infrastructure and what are some of the requirements that they’re asking for that’s a bit different these days?
ANDY:
So I think we’ve seen an uptick in enquiries, for sure. There’s definitely a greater skew towards cloud, we’re a dual strategy organization: we provide space and power and infrastructure for enterprises who want to host their equipment, but we’re also set up to provide the compute as a service. So we’ve seen an increase in enquiries, people asking about the source of the energy and usually they’re a little surprised when you explain that not only is it 100%, renewable, it’s all hydro, but it’s actually cheaper as well. People are used to kind of paying a premium for that. And I think there’s a greater awareness of some of the downsides of the hyperscale, or the big cloud players. It’s sort of like that old embrace and pollute approach to standards where large industry players would kind of add features, and then you’d be locked into their ecosystem. So if you’re working with the largest cloud provider that are out there, you know, you have to kind of craft your code and craft your site to sit happily in that environment and then you’re locked into high transfer costs, and so on. And we’ve taken a much more open view, in line with our OCP community membership and being an OCP ready facility, that we should have an open view to what you run as well. So we’ve tried to work to make sure that our platform – you can basically run anything on it – you’re not locked in with any kind of proprietary requirements for those VMs.
How would you prioritize the areas of efficiency when you’re speaking with an enterprise customer?
STEVE:
I would say that you guys have been extremely open, both about your intentions and your designs, and then sharing how that experience has been from the time that you’ve built it to the time that you’re rolling in racks at pace which has been really beneficial for the community.
When you look at some of the customers, most of these customers always say that they’re going to be needing something really efficient and that data centers will focus on their efficiency. So if you’re an enterprise customer, efficiency is a very broad topic, it could be at the IT level, it could be at the facility level. How would you prioritize the areas of efficiency when you’re speaking with an enterprise customer?
ANDY:
So the headline efficiency that people talk about is PUE, power usage efficiency, which is a measure of how much energy is used in cooling over and above the IT load. So a theoretically perfect data center that was using any energy for cooling at all, we have a PUE of 1.
Now, conventional DCs which spin hot air around and around using indirect cooling can be running typically at 1.4, 1.5 PUE, they’d be as low as 1.3. But that means for every kilowatt that you’re using, where every hundred kilowatts you’re using, you’re wasting 30,40, 50 kilowatts just on energy. And that’s because you’re having to take the same heat air around the loop and cool it down every time. Whereas with the novel approach to that we’ve used and that Facebook have used, we get right down to a PUE of 1.07, which is hugely better and means that you’re not wasting energy on cooling.
Water usage efficiency is important too. I know, Paul, that you’ve done some analysis on that?
PAUL:
Yeah, we’re about 30 times more efficient in terms of water that we don’t use, if you include water used in the cooling of the data center, as well as water not used in the generation of electricity, some even renewable energy sources can be extremely water-intensive, or non-fossil sources specifically, nuclear, can be extremely water intensive. So when you look at the overall picture, we’re saving a lot of water due to the way that we’re using direct free air cooling.
ANDY:
That’s right. And there’s a couple of other areas of efficiency that also people don’t always think about, one of which, when you think about OCP equipment is the efficiencies that you get in the ease of servicing the hardware for your engineers, having everything at the front of the rack, easily accessible, is a big improvement over a lot of server designs. And then also the power design in OCP racks. Having those DC bus bars and then the integrated power shelves, gives you a much greater efficiency between the power socket and the server. And that’s something that, while it doesn’t show in a PUE calculation, definitely plays into improvements in the overall efficiency of an infrastructure.
Can you talk a little more about the new cloud services you are launching?
STEVE:
That’s great. So it’s not just the efficiency at the facility level, but then all the way through the IT layer and the way that you’ve proved that. Those numbers, Paul, are really impressive at 30 times more efficient. And I really also like the way that you’ve expanded some of your services around OCP and into the cloud. So, do you mind talking a little bit about your new cloud services that you are launching?
PAUL:
Yes, sure. Our cloud product is really born from the need to offer a really strong alternative to the large public clouds that are available today. Clearly, a lot of those are very walled-garden type experiences and our clients are using them are obviously benefiting from large scale efficiencies from closing their on premise data centers and moving to the cloud. But having said that, there’s still a need for an open alternative, where there’s a big cost advantage, there’s an openness to moving data on and off the platform, and not being locked into using vendor-specific tools to do so. And then of course, there’s the data sovereignty aspect of being located in the EU, as compared to using a large American provider. So we see and saw an opportunity to offer something quite a bit different in the marketplace, where our customers can take advantage of openness, something that’s simple to use, and as Andy alluded to earlier, there is an increasing focus on the green aspect of data center and cloud usage.
Consumers are driving this for now. But ultimately, businesses are also looking at extremely closely at the carbon footprint of their operations and it can be quite tricky to get your cloud provider or your colocation data center operator to be transparent about carbon footprint, and that’s something that Hydro66 have done a pretty good job on. It is being transparent about the source of our energy, and helping our customers on the journey towards net zero possibly.
ANDY:
So I’m just going to add that I guess there’s another aspect to the platform that we have, which is we do see people liking or wanting to do multi-site or multi-region as a resilience approach, as opposed to the old school kind of trying to build up a very, very almost over provisioned single site where there’s things that can go wrong. So our cloud partners that we’re working with have other regions that we work closely with so enables us to deploy services in those regions too, while still being sure that if you specify that a particular service is to be in Sweden, and to run in a particular way, it stays there. So it’s almost the best of both possible worlds being able to have a very clear idea, and guarantee the location and the sovereignty of your equipment and data, but also be able to seamlessly add in extra regions, rather than trying to build them on competing public cloud infrastructures.
STEVE:
Well, that’s great. I didn’t realize that you had that type of control in the way that you’re rolling out services. And you have also your transparent and your decarbonisation efforts expand into the IT because I know you have relationships with some of our circular economy providers, but you’re also providing new hardware as well.
Over the next 6 to 12 months, what is on the horizon for Hydro66?
As you look at your strategy over the next coming months, in the next six to 12 months, what is on the horizon for Hydro66? Are you expanding the current facility up there? Are you looking at additional facilities? What’s in the plans?
ANDY:
Yeah, I mean, Steve, you’re absolutely right to say that we’re excited by the circular economy because we see real carbon and cost savings available in that space by taking equipment that has still got plenty of life in it, and that is being used very, very close to our main site, and giving it a second lease of life and also being very easy for clients and partners to deploy very quickly large amounts of infrastructure at rack scale. In fact, if people jump onto the website, they can see a video of a rack turning up being unpacked and slotted in in a matter of minutes.
In terms of what’s coming up for us, I think we’re going to see new cloud features. There’s stuff that we can’t necessarily quite announce yet but I’m pretty excited by the pipeline of what we see. We’re also building in some extra capability in terms of the GPU space, and machine learning and AI, and inference and training all the stuff that you can do around that and there’s some amazing new hardware that’s coming out. And we have some early contracts for that we’ve already announced. And so I see more growth of that kind of customer infrastructure.
In general, pretty optimistic on the pipeline, I think people are getting woken up to the importance of hosting in a green way and hosting in an efficient way and the future’s bright as far as I can see.
STEVE:
That’s great, Paul?
PAUL:
Absolutely. I would completely echo what Andy has just described. We’re glad that society in general is moving towards a more conscious approach towards the footprint of operations. Data center and data handling in general and storage is relatively easy to solve – it is a big challenge and a large amount of data that’s growing like crazy as we all know, but it is relatively easy to solve. There are solutions available today. Not just Hydro66 but other companies as well. I’ve been at this for some time, now. So there are solutions for forward thinking companies to go ahead and deploy infrastructure in the right environment in the right place. If you’re moving to the cloud as well, there are solutions available to take advantage of open simple public cloud services that can make it really easy for you to be transparent about your environmental ambitions.
So we are absolutely excited about the way that this is going to play out not just for Hydro66 but for the planet in general. We know that this story is not going to go away and we aim to be in the front for providing solutions for it.
STEVE:
That’s great. I know that Hydro66 is very well known in Sweden, and in Luleå itself, but it’s getting to be really well known not only throughout Europe, but also definitely throughout the OCP community. Really appreciate all the leadership that you guys have done in jumping on early to the OCP Ready strategy, and making this kind of a cornerstone for your efforts moving forward. Paul, Andy, thank you very much for the time today, I’ll be back with a couple of the big takeaways from my conversation with Hydro66. Thanks.
PAUL:
Thank you.
ANDY:
Thanks Steve.
Takeaways from the conversation
STEVE:
I’d like to share just a couple of takeaways from my conversation today with Paul and Andy from Hydro66. The first takeaway is how they took a look at what Facebook was doing around their designs and their location worldwide and took those ideas and those best practices and localised them to start a new business serving Europe, North America and Asia through Hydro66.
The second is their take on efficiency, and I found this discussion the most interesting, the cheaper renewables that you get out of Sweden and one of Paul’s stats of 30 times more efficient in the water that they don’t use in that facility, and how that compares to the balance of Europe and their overall power prices. It’s not just the efficiency at that level but it’s also the efficiency that they’re driving at the IT level as well, and that brings me to my third takeaway: is how they’re embracing circular economy. They’re one of the early adopters of circular economy, OCP gear within the Nordics and how they’re building cloud services now around those circular economy offerings. It’s going to make a real impact in the adoption across Europe and the other regions.
So again, if you have any questions around Hydro66’s facility, please reach out to myself Steve at opencompute.org, I’m happy to introduce you to Andy and Paul so they can give you a virtual tour of their facility in Luleå, Sweden. It’s quite impressive.