Page 8 - Spring 2025 Electron

t
AI is seen as complementary to the above through
its ability to introduce intelligence and adaptability
into operations. The ability to train on richer datasets
derived from quantum simulations is expected to
accelerate breakthroughs in areas such as natural
language processing, image recognition and
predictive modelling:
‘ Integrating quantum computing and AI into
data centre architectures represents a profound
technological shift with immense potential to
address long-standing challenges. While both
technologies are powerful individually, their
combination enables synergies that unlock
entirely new capabilities.’
Reference: ‘Could Quantum Computing and AI spark the next
Revolution in Data Centres?’, Data Centre Review, 13th. February
Addressing the Hyperscale Data Centre
Bandwidth and Energy Consumption Crunch
Hyperscale data centres are increasingly requiring
higher bandwidth and lower latency in order to keep
pace with modern data processing demands.
These requirements have prompted Lessengers Inc.,
a provider of innovative optical components based
on its patented ‘Direct Optical Wiring’ technology,
to launch its industry-first 1.6T OSFP 2xSR4 optical
transceiver with multimode optics to address the
growing bandwidth and energy crunch resulting from
the increasingly complex AI/ML workloads that now
characterise hyperscale data centres.
Direct Optical Wiring technology is a polymer-
based air-cladded waveguide technology that is
particularly useful for optical interconnects in data
centres and other high-performance computing
environments. It is a direct coupling method that
replaces the use of lens optics in transceivers,
which can introduce a variety of errors and optical
impairments that have a negative impact on signal
quality, transmission efficiency and overall system
performance.
Dr. Vladimir Kozlov, CEO and Founder of LightCounting
Market Research, comments:
“ For AI/ML workloads in hyperscale data centres,
LPO and LRO are increasingly favoured for their
power efficiency and low latency, whereas retimed
optics are used when the link accountability and
interoperability are prioritised. Therefore, not only
are we seeing the progression from 800G to 1.6T
optics, but the transceiver suppliers must offer a
range of power budgets as well.”
Article supplied courtesy of LouVan Communications Inc.
8 9
Proof-of-Concept for Orbital Data Centre
Technology
Axiom Space’s Data Centre Unit-1 (AxDCU-1) has been
confirmed as representing the company’s initial proof-
of-concept for orbital data centre technology. This
smaller prototype will process data in real-time on the
International Space Station prior to the larger Axiom
Space T1 data centre deployment, due to arrive in orbit
in 2027.
It is expected that these data centres will offer
on-orbit data processing, moving large-scale
computing closer to where it is produced, such as
satellites or spacecraft, reducing reliance on ground-
based facilities. Other applications include potentially
AI/ML training, off-planet backup and disaster
recovery, faster data relay via optical inter-satellite
links, multi-factor authentication, and cyber intrusion
detection and response.
At the heart of the AxDCU-1 will be Red Hat Device
Edge, which leverages the Red Hat-led community
project MicroShift, a lightweight Kubernetes
distribution derived from Red Hat OpenShift, which
is supported by Red Hat Enterprise Linux and the Red
Hat Ansible Automation Platform. These tools are
designed to enable cloud-native applications on-
orbit.
Tony James, Chief Architect, Science and Space, for
Red Hat, states:
“ Off-planet data processing is the next frontier and
edge computing is a crucial component. With Red
Hat Device Edge and in collaboration with Axiom
Space, Earth-based mission partners will have the
capabilities necessary to make real-time decisions
in space with greater reliability and consistency.”
Jason Aspiotis, Global Director of In-space Data and
Security for Axiom Space, adds:
“ Infusing terrestrial-grade cloud solutions
into orbital data centres will enable users to
seamlessly transition and enhance their terrestrial
workloads to orbit while leveraging the lower
latency and increased security inherent with
orbital data centres.”
Reference: ‘Red Hat partners with Axiom Space to bring Orbital Data
Centre to Life’, Data Centre Review, 6th. March
Become a member: https://membermojo.co.uk/ie
Become a subscriber: https://membermojo.co.uk/subscriber
Become a sponsor: https://membermojo.co.uk/sponsor
DATA CENTRE CASE
STUDY: VIRGIN
MEDIA O2
Virgin Media O2 was launched on 1st. June 2021,
combining the UK’s largest mobile network with a
fully gigabit broadband network. As it rolls out its
next generation of 5G mobile and fibre networks, an
increasing workload demand is being exerted on its
network of data centres.
The Technical Site Engineering and Delivery Team
has the task of optimising the energy performance
of these data centres. For this EkkoSense, AI-powered
data centre optimisation specialist and winner of
the 2025 Data Centre Awards Special Contribution to
improving Energy Efficiency category, was engaged
to obtain a real-time view of the thermal power and
capacity performance across Virgin Media O2’s
twenty UK data centre sites.
EkkoSense differs from the traditional IT-led DCIM
approach in that it offers a distinctive, light-touch
AI-enabled software-driven thermal, power and
capacity optimisation solution, enabling operations
teams to optimise data centre performance whilst at
the same time delivering quantifiable sustainability
results. Low cost IoT sensors provide the innovative
Software-as-a-Service (SaaS) platform with valuable
data, while machine learning provides AI analytics,
gaming-class 3D visualisation and Digital Twin
capabilities.
Granular data centre sensing provides the core
machine learning data that enables true real-
time visibility of cooling, power and capability
performance, with potential cooling issues being
able to be identified quickly using comprehensive 3D
Digital Twin visualisations that allow information to be
monitored and interpreted quickly.
Attaching thermal sensors on all racks and cooling
systems allows for the capture of granular and
continuous thermal data, which is fed into the EkkoSense
software while unique EkkoAir Wireless cooling duty
sensors additionally provide real-time tracking of cooling
duty loads in kWc for any CRAC/AHU unit.
Become a member: https://membermojo.co.uk/ie
Become a subscriber: https://membermojo.co.uk/subscriber
Become a sponsor: https://membermojo.co.uk/sponsor
s
ndu
s
e
w
r
y N
I
Industry News
With the EkkopSense software collecting thousands
of data points every five minutes, the millions of data
points already collected contribute directly to the
effectiveness of machine learning algorithms. Ai and
machine learning therefore scale up optimisation
opportunity enabling complex data centre datasets
to be processed in seconds where weeks would have
previously been required.
The installation of sensors, deployment of EkkoSense
Critical software and optimisation of thermal
performance took an average of two weeks per site,
with the cooling performance of all of the sites being
optimised in six months. Cooling energy savings
across the sites has amounted to £1 million a year
representing an average saving in data centre
cooling across the sites of 15 per cent. This energy
saving is equivalent to 760 tonnes of carbon dioxide
using location-based scope 2 accounting.
The project has also resulted in improved
performance from Virgin Media O2’s significant
investment in free air cooling devices by maximising
free air capacity. Where possible, reducing maximum
rack temperatures has also helped to resolve
potential thermal risks associated with hot spots.
Adrian Lazenby, Head of Technical Site Engineering
and Delivery for Virgin Media O2, states:
“ The good news for Virgin Media O2 is that
EkkoSense continues to add value to its AI-
powered 3D visualisation and analytics software
solution. That’s contributing with innovations
such as Cooling Anomaly Detection that will help
us identify machinal and electrical equipment
performance anomalies before potential failures -
and we’re looking forward further functionality as
the software continues its development. “
Article supplied courtesy of EkkoSense