THE EDGE
DATA CENTRE
AND COLOCATION
PROVIDERS MUST
HAVE A SINGLE
VIEW INTO THE
FACILITY AND
THE DATA THEY
ARE HOLDING.
Artificial Intelligence – a new way
to apply critical thinking
AI is beginning to power some DCIM
solutions in a few key areas:
Targeting: Data centre and colocation
providers must have a single view into the
facility and the data they are holding. Such
a granular level of visibility requires DCIM
and there are smart ways that AI can
make this process smoother, by making
decisions and actions for the human user
to get to an optimal state quicker. Such a
solution will address data centre analytics
by collecting, normalising and creating
patterns of facilities and IT data and
streaming it to a control centre.
The solution then uses its Machine
Learning (ML) capabilities to extract
predictive models to send the analysis
back to a visual dashboard to display the
potential vulnerabilities, such as future hot
server rows or under-utilised racks.
What’s the difference between Machine
Learning and AI? Machine Learning is the
part that looks for patterns and AI is the
automatic decision maker that acts on
what is identified within these patterns.
Often it is sensible to refer to both
together as part of the same solution, e.g.
ML/AI. Together they can help with:
Collecting: Capturing data from all
distributed silos such as servers, sensors,
HVAC, building monitoring software,
PDUs, processors and many other points.
Analysing: Advanced content analytics
enable facility managers to understand not
just what happened, but also how and why.
70
Issue 12
Actioning: Refining data into a visual
state so team members may quickly
comprehend current conditions as well
as increase operational efficiencies and
cost savings.
Efficiency: DCIM not only allows
facility managers to target the data they
hold, it can also help mine additional
environmental data to help run the
facility better and more efficiently. This
is the reason why many data centre
and colocation providers are turning to
DCIM powered by ML/AI for the everyday
running of their facilities.
The technology can improve efficiency
by using algorithms which collect data
from thousands of sensors all around the
network which in turn feed into an ML/AI
system that is modelled on neurons found
in the human brain. The system then
analyses the broad range of indicators,
from energy consumption levels to safety
constraints, in order to identify the best
course of action.
This makes the process more unified
as ML/AI is creating a cohesive
platform for the infrastructure strategy
by incorporating the processes, tools
and workforce focusing on end-to-
end solutions, allowing initiatives to
work together by design to reduce
duplicate effort.
For example, by recording the airflow,
the system can identify if any of the air
filters are clogged, then notify the team
and in turn push the air through less
clogged filters until they are changed.
Once changed, the system would resume
service as usual.
Security: ML/AI can be constantly
monitoring every part of the network. This
gives managers the means of dealing with
the growing influx of data while learning
and adapting to overcome new, never-
seen-before malware while recognising
suspicious user behaviours and detecting
anomalous network traffic.
It can also collect and analyse forensic
data, scan code and infrastructure for
vulnerabilities, potential weaknesses and
configuration errors, making it one of the
most powerful tools at security’s disposal.
Implementing such ML/AI solutions
with DCIM can reduce reliance on
human intervention by reducing the
man-hours spent on round-the-clock
monitoring and decreasing the risk of
human error in response.
The future is just around
the corner
A future of ML/AI and AR everything
is around the corner but needs to be
carefully applied so that the enormous
investments in data centre uptime
are maintained.
AR is a non-disruptive technology that
augments the human’s ability to work
smarter and faster.
ML/AI is a slightly trickier concept – and
may well be applied in smaller projects as
the industry learns just how best to use
it. It is easy to imagine broader use cases
emerging as successes snowball – and
if that occurs those who lead in early
adoption may find massive gains put them
well ahead of their peers. ◊
www.intelligentdatacentres.com