IXcellerate offers customized colocation solutions for high-load IT infrastructures.
We always take into account the necessary technical parameters and requirements to fulfill SLA and minimize the accident risks. Our in-house expertise and proven engineering solutions ensure the uninterrupted operation of your IT infrastructure.
Ready-made solutions for
high-load colocation up to 55 kW per rack
Total power 380 MW
Dedicated data halls for supercomputers and AI-based solutions
Scalability colocation of 1 to
over 1000 racks
Real-time monitoring of infrastructure and perimeter security
Unified network circuit (interconnects)
Effective heat dissipation
through free cooling and low-speed ventilation (LSV)
Readiness for elevated air temperatures from
servers’ outlet
Roadmap for successful launch in the Data Center
Infrastructure for
high-performance
computing
topology
High-performance servers require numerous network links and interconnects, thus the rack colocation in the data hall differs from a traditional row system: the distance between racks varies depending on the load (supercomputer) and functionality (switching).
system
When designing high-density computing, the heat output of the entire data hall is considered to prevent overheating of specific zones. Depending on technical solutions (based on raised floors or “cool walls”) the width of cold and hot aisles can be increased based on the required air volume.
Solutions utilizing low-speed air circulation (LSV) enable more efficient cooling of servers without the need for inter-rack air conditioning units. IXcellerate supports autonomous cooling supply for engineering systems through free cooling and adiabatic processes.
scheme
Within a single rack, multiple blade chassis and disk shelves can be installed without the need for inter-rack connections. This scheme simplifies the construction of a computing node and its further operation while reducing costs for network services, structured cabling systems (SCS), and infrastructure maintenance.
-for cooling
power supply
Reviews
Give us a rating
Give us a rating
Pavel Shchegolev
18.12.2023
Give us a rating
Oleg Ionenko
10.10.2023
Still in doubt?
This trend is associated with new digital realities: the daily launch of new digital projects, the growing popularity of artificial intelligence and neural networks, and the increase in data volume. Speed and 24/7 access have become the main quality indicators for online services.
The performance of computing is increasing and requires more powerful processors and servers. Infrastructure built for classical computing environments cannot provide the necessary power supply and cooling for next-generation servers.
The maximum amount of heat that a modern chip’s cooling system (Thermal Design Power) must dissipate reaches 300 watts. Processor manufacturers are announcing plans to increase thermal power to 400+ watts per chip in the next 3-5 years. To maintain the temperature regime in a standard rack (5-7 kW), it is possible to install no more than 7-10 servers with next-generation processors, and to maintain the temperature regime in the machine room, racks must be spaced apart.
To ensure ultra-fast access to shared memory in such systems, many network connections are required, which increases costs for customers in organizing structured cabling systems (SCS) and managing IT infrastructure, complicating the process of upgrading platforms to more powerful and energy-intensive solutions.
They can, but at the cost of usable space in the data hall. High-performance servers require a greater number of network links and interconnects; therefore, placing high-load racks in the machine room differs from a classic row system: the distance between racks varies depending on the load (supercomputer) and functionality (switching). High-load racks and standard racks should not be placed next to each other or directly opposite one another to avoid a shortage of cold air intake.
First of all, it is essential to clarify whether the data center has infrastructure with guaranteed power supply up to 55 kW per rack and whether it can ensure the uninterrupted operation of your IT platforms in the long term.
What is the available power capacity of the data center itself? In other words, does it have the capability to support the scaling of already hosted resources and the launch of new digital projects? The declared capacity should be sufficient not only to support customer equipment but also to ensure the uninterrupted operation of its own engineering infrastructure in accordance with energy efficiency indicators (PUE).
It is also important to pay attention to the load distribution scheme, redundancy scheme, and other parameters of the data center that directly affect its reliability and fault tolerance level.
Semyon B.
25.11.2021