17 June 2021
Edge and the micro data centre have become essential elements in the expanding data network that is the platform for our data economy.
Gains in processor performance and the need for location specific compute for the myriad applications that require ultra-low latency communication, means that features that were based in traditional data centres are now deployed where the user and the data source reside.
The micro data centre is a means for processing Edge data, which may be gathered from vehicles, traffic automation systems, surveillance devices, local 5G networks, office and factory facilities.
The data must be gathered and processed locally to improve the performance of the overall system.
As the volume of data increases so does the need for micro data centres’ performance to rise to utilise the AI and ML-based applications required to process, analyse and respond to the data.
Today’s micro data centres are standalone systems such as the Ku:l Micro Data Centre, offering highly configurable and scalable solutions to enable accelerated corporate applications, HPC (high performance computing) and AI workloads in any Edge environment whilst maintaining data centre density.
These micro data centres can be as small as a filing cabinet and scale up to 48U rack cabinets and are often usually configured for specific workloads to maximise efficiency.
As standalone systems they contain features typical in traditional data centres: CPU and GPUs, a cooling system, storage, security and reserve power source.
Iceotope’s Ku:l Micro Data Centre utilises Lenovo’s ThinkSystem SR670 2u rack server and delivers HPC performance supported by four GPUs per server.
Integrating the server into a liquid cooled chassis eliminates the need for any air cooling, delivering an almost silent, energy efficient solution with >95% heat capture and rejection via a heat rejection unit (HRU) and the sealed chassis removes the need for humidity monitoring. iceotope.com