ClusterPower: a Green Hyperscale DC Set To Dominate Sector
“We have just become an alternative to the entire FLAP area,” says Vladimir Ester, CTO of ClusterPower, as they position themselves to become one of the most groundbreaking DC centres in Central and Eastern Europe.
The FLAP areas are Frankfurt, London, Amsterdam and Paris — the four cities considered to be the primary markets for all new data centre deployments.
Located in Romania, ClusterPower’s five hyperscale Data Centres (DCs) are of a magnitude and scalability that promises to see them become the leading provider for the entire region, while enabling them to tap into international markets. All five data centres are scheduled for completion by 2025, by which time they will be housed in a 25,000 SQM campus, sustainably producing a total capacity of 200MW.
So this is hyperscale technology — but not as you know it.
Ester comes from a strong technological background and has acted as an infrastructure engineer, as well as a solutions architect, for data centre and internet service provider deployments throughout most of his career. He has also been a CCIE — Cisco Certified Internetwork Expert — for almost 10 years, with this expertise laying the technical foundations for project ClusterPower. “I think that this skill set has really prepared me to both design and implement large-scale infrastructure projects,” he says. “So, in the last couple of years, I've moved away from engineering towards management and entrepreneurship, joining Cosmin and the rest of the team, and I’m now managing all of the infrastructure aspects related to our campus.”
“We are very experienced in our fields,” says Cosmin Georgescu, CEO of ClusterPower, “and I have been closely involved in both energy and technology projects for over 20 years now, so it was a natural move to have our own assets and to build our own Data Centre and technology campus.”
ClusterPower is the first ever Hyperscale Data Centre in Central and Eastern Europe; both the technological and operational vision behind it is just as huge.
Georgescu possesses a fundamental knowledge of Cisco training and certifications, as well as for Project Management, with these experiences leading to the development of the first data centre in Romania for Microsoft, Bosch, and Amazon. He’s also worked on several large national projects and infrastructures, including establishing a nuclear power plant, Romania's first wind farm, and its first photovoltaic plant.
Speaking about Clusterpower’s capabilities he says: “We can host large scale infrastructure deployments in a custom, fit-out way. We are very flexible in customising services for our clients. We can host for clients and customers that either want to expand in the Central Eastern European region, or that want to optimise their current infrastructure.”
A Green Energy Data Centre Powerhouse
One of the main cross-industry concerns with hyperscale DCs is potential energy wastage. But, through strategic partnerships, ClusterPower has devised an innovative solution: build a powerful data park that is not only operationally effective, but also highly sustainable and extremely energy efficient.
“We are focusing on green energy, power and efficiency,” says Georgescu, and that focus is already yielding major returns. ClusterPower has a PUE (Power Usage Effectiveness) of just 1.1, which is considered very low across the industry. “We produce our own power and cooling at the same time and, to come up with such an innovative solution, we worked with our partners, Rolls-Royce MTU (hydrogen-engine solutions), which makes us not only extremely efficient but also allows us to function independently, while greatly reducing business interruption risks.”
Continuity of service is absolutely critical to DC hosting, since entire data infrastructures depend on their uninterrupted operations.
“We already have eight redundant fibre connections on site,” Georgescu says, “and that’s just in the first phase.” Redundant connections are essentially duplicate databases that are ready for access in the event of unforeseen circumstances.
“We have also established the base of our digital highway connections to Frankfurt and Amsterdam. ClusterPower’s infrastructure already provides our clients with hundreds of racks, as well as off-PCs for 200 people that may be used to resolve business interruptions.”
This means that, if their clients have a failover in a large European city, they can take a flight to nearby international airports — which are just 20 minutes away from ClusterPower’s base — and continue to work directly in their data centres with 20 or 40 people without any interruption, depending on their use-case.
ClusterPower is a real powerhouse of a Data Centre. Its sheer output of 200MW will allow it to produce its own sustainable electricity in-line with EU regulations, from its strategically positioned location.
Georgescu says: “One important aspect of our project is being located in the EU, which means that we follow all European security and data privacy guidelines and, geographically speaking, we are at the gateway between East, West, North and South. This gateway is positioned to reach and supply a multitude of cities that can in turn reach us and our populations here.”
ClusterPower’s geographic placement is a decisive point of client-attraction. He says that the company is “in an area with very low seismic activity for about an 800 km radius”, providing both security and critical dependability as a direct consequence of its site location.
Core Principles of Security and Resilience
Security and resilience are core principles of ClusterPower, as well as the design and digital architecture being built entirely around them. ClusterPower’s campus is protected according to military standards and continuously monitored by over 200 high-definition cameras. All access is controlled by specific security procedures, with individual security levels for each rack, row and room — including biometric access technology.
ClusterPower’s security-first model can be divided into two areas: the first is related to specialised cyber security equipment, such as next-generation firewalls and its own distributed denial of service protection (DDOS), scrubbing centre and web applications; the second area is related to systems’ resilience, which provides a powerful infrastructure that will remain operational even during an attack or disaster.
All of ClusterPower’s implemented solutions are fully redundant and, in terms of N+N — database duplicates that effectively increase data transfer speeds — the company has spared no expense. This means that, according to its own calculations, ClusterPower surpasses even the Tier-Three requirements of an uptime of 99.985%. And Georgescu believes the company can exceed even that.
All of these cutting-edge components and systems render ClusterPower a truly formidable competitor in the Data Centre industry.
“We see ourselves as the gold standard,”’ says Ester. “From physical infrastructure, connectivity, utilities, resiliency — all the way up to the most advanced IT and technological appliances in our DC — everything has been conceived as state-of-the -art, even down to the procurement process, where we have chosen to work only with industry leaders for their segments.”
“All of the networking infrastructure is being deployed in a partnership with Cisco. We use Paolo Alto and F5, who are leaders in security and low-balancing; we're using NetApp and grid storage solutions; and, last but not least, we work with NVIDIA and the our co-created AI platform, which is indeed a gold standard all on its own in the AI industry.”
Artificial Intelligence and Supercomputing Data
ClusterPower firmly believes that machine learning and artificial intelligence applications are here to stay, as they present both a huge market opportunity and a decisive direction for the industry to move in, as a whole.
Ester says: “This is very closely related to other technologies, such as twin digital ecosystems like metaverse. This all relates to AI. Clusterpower has decided to fill in the gap of being the first cloud provider to offer AI as a service in this region, and we have built this service alongside NVIDIA, which means that we have installed Romania’s first supercomputer based on the DGX platform from NVIDIA, alongside Mellanox Infiniband Switch and NetApp storage.”
In a nutshell, this is a certified supercomputer that ClusterPower is offering as a service to its customers who want to either train their own AI models or run their AI-based applications in.
“So we fit both models with the same infrastructure, research, training and inferencing,” says Ester. “This is a very scalable and flexible service model that we have designed, and I think that with this kind of service offering, we are pioneers in the field — not just the region.”
Georgescu reaffirms this perspective: “It provides us with a very powerful tool for management and cost optimisation, as well as for predictive maintenance. This has already been implemented on our campus, where we’re using artificial intelligence for energy optimisation as well as forecasting.
“Most importantly, we have a great ecosystem of partners and a very seasoned team,” says Georgescu. “Together, we all share a vision to promote a project that will change the world for the better — which means that, not only do we aim to support our customers to achieve their most ambitious goals, but we can help them become part of a transformational process, such as large-scale use of trust policy AI, which will benefit regional developments in making them both competitive and unique, while allowing them to use state-of-the-art tools.”