final thought
PRESSURE TO PERFORM
Adrian Barker of RF Code highlights the need for a reliable data centre.
T
he amount of data
created each day is now
at an astronomical 2.5
quintillion bytes. What is
even more astonishing is
that 90 per cent of the world’s data
was created in the last two years.
This data is created by our use of
social media, streaming films online,
Internet banking – almost every daily
task involves data.
The increase in data creation
in such a short space of time
means the data centres powering
this growth are under increasing
pressure to perform. Data centre
downtime has always been costly but
even more so now daily tasks are
so dependent on an uninterrupted
service from these facilities.
The move to the cloud and the
impending boom of connected
devices and sensors to power the
Internet of Things means our need for
reliable data centre performance is
greater than ever.
The damage of downtime
The cost of data centre downtime
is more than just the financial
implications of fixing failed equipment.
The damage to brand value,
reputation, business and productivity
can be just as substantial.
Delta Air Lines’ outage in August
is reported to have cost $150m,
including customer refunds for
48
cancelled and delayed flights. The data
centre outage was only for five hours
but this was enough time to cause
disruption to 2,000 flights and damage
the company’s revenue and reputation.
Delta had initially stated the issue
was due to a power cut but Delta’s
CEO later revealed the cause was
a lack of back up power for some
crucial servers. With organisations
increasingly creating complex hybrid
IT environments by combining owneroperated data centres, colocation
and cloud technologies, there is
unfortunately more room for error.
The latest Ponemon Institute
survey revealed that there has been
a 38 per cent increase in downtime
costs since 2010 and maximum
downtime costs have increased
by 81 per cent - as data centres
become more business critical, a
data centre failure has become
even more expensive. This is due
to hardware costs, data loss, legal
costs and the negative effects to the
brand’s reputation.
And with data creation continuing
to grow, the costs of outages will
likely continue to increase.
More data, more problems
There are many factors that can
affect data centre and service
availability. For example, a spike in
data centre traffic due to a peak
shopping period or cultural event
can cause data centre equipment
to work overtime. This can cause a
large fluctuation in the thermal output
of equipment which can cause even
previously effective cooling systems
to become drastically overburdened.
Another effect is overheating if
efficient cooling is not in place, and
if this scenario continues over a
prolonged period of time, equipment
will be damaged or fail. The same
outcome can occur if there is a
lack of capacity available: demand
for service suddenly increases, but
without sufficient availability the data
centre is unable to deliver.
Just look at what Pokémon Go’s
developer Niantic’s expectations
were for the game’s server load
against the reality – the company
was 50 times out. This meant a
lot of disgruntled players and a lot
of downtime. Luckily the game’s
popularity was so great that people
were willing to persevere but it could
have very easily caused many people
to stop using the App altogether.
That said, data centre
managers need to be cautious of
overprovisioning and the purchasing
of unnecessary services. Working
with a hybrid IT environment can
easily lead to overprovisioning if
there are no up to date records of
capacity and equipment across all
facilities. It has been known for
companies to have boxes of brand