The GreenCloud Unit enables sustainable local cloud, providing the cost efficiency, edge capabilities and privacy of local compute; and the resilience, scalability and full outsourcing of maintenance of hard-and-software of the cloud.
Turn your existing servers into useful warm water generating machines. We make you existing servers immersion ready and compatible with the Cloud Unit.
Optimized for the highest performance and outlet water temperature.
Above the ground If you need temporary high-performance computing, the Cloud Unit can be placed anywhere. It can be operational within minutes.
In the ground - Ideal for situations where continuous heat is required year-round. Placing the Cloud Unit into the ground makes it more efficient and secure. Ideal for offices, residential housing, industrial processes, swimming pool or aquatic ponds or laundry facilities.
A Cloud Unit has two casettes. Each casette can has one of the follwing flavors: compute, storage or AI.
Good balance between the performance and the available cores per server. Ideal for compute intensive tasks like FaaS or simulations.
Cloud Unit (two cassettes):
* Physical space is not our limiting factor. Our CPU cores perform 30% faster than the high density 64 or 128 core CPUs.
Heating capacity:
We can use a combination of NVMe, SSD or helium HDDs.
Cloud Unit (two cassettes):
The field of AI is advancing rapidly, please contact us to discuss your needs.
In 12 steps
We design the Cloud Unit from scratch, allowing us to have full control over its properties and to iteratively optimize the unit for increased performance.
The stainless steel parts are precisely laser cut while other components are ordered simultaneously.
The stainless steel parts are TIG welded together to create a waterproof container. This container houses all the components typically found in a state-of-the-art data center.
The waterproof stainless steel container is wrapped in PIR insulation to minimize heat loss and is then encased in reinforced concrete to prevent unauthorized access. The lid is separately poured, four times thicker, and weighs over 150 kilograms.
We design and build our own electronics to control the actuators (pumps, motors, valves, etc.) and to collect sensor data (vibration, temperature, humidity, flow rate, cameras, etc.).
Off-the-shelf servers cannot run efficiently in immersion fluid without modifications. For example, thermal paste typically dissolves in immersion liquid, which is resolved by replacing it with graphite sheets or indium metal. We also install our unique solutions to maximize heat extraction from the servers.
The Out of Band system (control systems), servers, and other components are installed into the Cloud Unit. The Out of Band system allows remote control of the servers, enabling power on/off, resets, and troubleshooting without opening the Cloud Unit.
The Cloud Unit comes with a base IT infrastructure, helping customers deploy their preferred OS, AI, or configure their storage on the servers.
With middleware deployment, we enable direct integration with cloud services like Azure HCI, AWS Outpost, or Azure Function Apps. Onsite integration with auxiliary systems such as robotics and 5G is also possible.
We integrate the Cloud Unit into the existing infrastructure of your building, providing localized heat or computing power where needed.
We offer a comprehensive Cloud Service, monitoring the Cloud Unit status to ensure maximum hardware performance and security against unauthorized access.
We are passionate about technology and solving challenges that impede your business operations. We are always available to help with brainstorming and implementing solutions.
If you want more technical details, then please continue to read further.
The default Cloud Unit configuration consists of two cassettes that contains each 2.5kW of compute power and a 1000 liter water buffer tank. Depending on the heat recipient use case we can deliver different temperatures.
In short, it depends on several factors:
Share your needs and we will calculate the optimal setup for you. We continuously monitor the efficiency of the Cloud Unit and adjust the parameters to achieve the highest efficiency (season, workload, customer needs etc).
The Cloud Unit will always be operational and you can use the heat of the Cloud Unit at any time. When the hot water is not used for a while (see graph above) then the maximum capacity of the "hot water buffer tank" will be reached. At this point the heat will be wasted and dissipated into the air. You will not be liable, it just ensures that the compute customers can compute on the Cloud Unit even if the heat cannot be reused.
If you want more technical details, then please continue to read further.
In general, any binary compiled for Linux or FreeBSD is supported. Besides running binaries we also provide FaaS instances with a Software Development Kit (SDK) for the following programming languages:
A simple POST API request can be used to create a FaaS instance. A JSON data structure is transmitted within the request body of the API request. The JSON data contains the parameters like the payload source URL, SDK version and datastore destination.
The task schedular ensures that FaaS instances are distributed equally between the servers.
The code that need to run on the FaaS instance can be downloaded from anywhere. And the result can also be uploaded anywhere.
If post-processing is needed by another FaaS instance its is recommeded to store the result on the local datastore within the Cloud Unit. The network bandwidth between the FaaS instances is 10Gbps (within the same Cloud Unit).
The commandline program is an convenient way to manage the Cloud Unit. It is a wrapper around CURL that uses API requests to communicate with the Cloud Unit. In this demo we will spin-up multiple instances that will execute a python script. The python script will collect all the links on the target website.