NETdepot, an IT infrastructure company came to us looking for a data center capable of delivering a truly uninterrupted power supply. Power was a real priority in this case, as the company in question designs, builds and maintains infrastructures for organizations all over the country.
The problem: A need for reliable power and scalable options
The client’s services required power at all times; interruptions would have meant a loss of quality. The team therefore needed a totally reliable option that would never require them to compromise in terms of the quality of their service.
In addition, the team was looking for scalability. The company’s ambitious growth plans would necessitate the purchasing of more equipment, and they needed a data center that would be able to accommodate that without any interruption to service.
The solution: A unique setup prioritizing power
Our team advised a system whereby the rack’s power would be delivered via six different cables, coming from six different power supplies. As you can see from the photo, it’s packed full of 8xH100 NVIDIA GPUs and has six cables coming from it.
We were able to deliver three unique feeds to the rack – which isn’t a common feat in the industry. Being able to do so is important, though, as it allows for a minimum of four unique power supplies from four different sources in the event of an incident. Therefore, there’s no real chance of an interruption to power or capability – no matter what happens.
In the near future, the client plans to purchase a very large multi-hundred GPU cluster with H200s, which the team will be landing and configuring with the same unique power configuration.
The end result: A future-proof solution that’s ready for growth
Scaling for the future was also important to this client. We worked with the team to come up with a solution that would allow them to grow, with the full support of their data center.
With this setup, cooling was particularly important. The rack generates nearly 10kW of heat off the back, so cooling systems had to be able to counteract that, and be reliable in doing so.
Systems like these are incredible in terms of what they can do, but there’s a huge amount of planning and organization that goes on behind the scenes to allow them to really fulfill their potential.
Few data centers have the tools and expertise required to facilitate such setups, but at TRG we’re fully equipped to look after such systems and really give our clients back that peace of mind. And when these systems are set up properly and looked after, with the right redundancy, they really do deliver great results.
