sk any operations team if their
data centre rooms are fully
performance-optimised and the
answer will invariably be yes.
However, despite the best efforts of FM and
IT teams, the reality is that even today’s
best run data centres have cooling, power
and space issues that should be resolved.
This is an industry-wide challenge,
particularly as powering data centre
cooling now accounts for around 30%
of a data centre’s overall operating cost.
And with research suggesting that just
a third of installed data centre cooling
equipment actually delivers active cooling
benefits, there’s clearly a requirement
for organisations to look again at their
data centre’s performance if they’re to
successfully remove thermal risk, achieve
cooling energy savings and release further
Unfortunately very few data centres have
access to the kind of real-time core power
and thermal metrics that they need to
make informed performance optimisation
decisions about their critical facilities.
Instead, many still rely on their original
CAD floor layouts to calculate power and
cooling requirements, while those that
do regularly collect thermal data often
consign it to inflexible spreadsheets that
are rarely either accessible or up-to-date
when critical information is needed.
This challenge was brought home last
month when I visited a data centre that
was undergoing a significant upgrade.
They were updating around 15% of their
existing racks and were adding 30 new
ones over the next few weeks. Given the
scale of change, how could they hope
to ensure the right thermal, power and
capacity balance if they had no access to
current performance data?
Getting serious about sensing
So, if you’re serious about optimising data
centre performance then you really need
to know what’s happening right now – not
what was going on yesterday or last week.
That’s why it’s only when data rooms are
carefully mapped with all the appropriate
data fields that operations teams can
start to gain a true understanding of their
overall data centre performance. When
a data room is carefully mapped with
appropriate thermal data fields – right
down to an individual rack or server level
– whole new levels of understanding and
cooling efficiency become possible.
To address this, organisations need to work
out how to build rack-level detailed maps of
their data centre estate that display all their
cooling, power and thermal performance in
real-time. It’s only by combining this kind
of granular cooling and thermal data with
TWINS MODEL AS
A GREAT WAY FOR
smart monitoring and analysis software
that they can access the intelligence
required to enable informed performance
optimisation decisions to be made.
Unfortunately, less than 5% of data
centres currently gather this kind of
precision data, as it requires a far greater
networked mesh of sensors to accurately
capture not just temperatures, but also
energy usage, heat outputs and airflows.
Turning critical data into
Given that so much data operator and
facilities management time is taken
This highlights a dilemma at the heart of
data centre optimisation. Do you choose
to lock down your initial data centre set-
up – in the hope of maintaining reliability
and performance? Or do you recognise
that IT will need to keep pace as your
operation evolves, requiring frequent
attention in order to keep pace with your
changing business requirements? The
answer should be clear, but delivering on it
will require true, real-time visibility of your
critical heartbeat operational data.