User Manual

> White Paper | Best Practices in Digital Transformation
22
ubusiness but also for clients by ensuring we maximise the use of
cloud systems that have proven to reduce costs and maintain high
service availability”. [Financial Services]
The scarcity of data related to the data center will no longer be the
characteristic that defines MTDC provider’s eorts to understand
their data centers. Increasingly the issue will be how to define what
constitutes useful information. It is estimated that by 2020 there will be
more than 200 billion sensors across more than 300 billion terminals,
each producing an average of 25 exabytes of data per month.
There are significant issues that accompany the deployment of
analytics and which determine its use in future decision making,
whether human or AI-based. These include issues of:
• Data generation: To deploy these big data/IoT initiatives,
organisations must first generate data. Although this might seem
unlikely, not all MTDCs generate sucient quantities of data
related to their data centers to make IoT analytics viable.
• Data quality: Decision-making requires data of a reliable
quality. Machines generate the virtually all the data in the
MTDC scenario, reducing potential quality problems. However
to ensure this, particularly in a situation of increased cyber-
vulnerability, analyses to validate data health should be deployed.
• Data curation: The potentially huge amounts of data that an IoT
initiative will generate needs curation – the process whereby what
is useful is distinguished from what is not. Without this step of the
process, the data generated runs the risk of flooding the networks,
processing and storage systems or at the least using considerable
resources unnecessarily. This is a core principle of edge computing
– that if processing can be done at the point where data is collected
or at nodes before the central processing activity, then this reduces
otherwise unmanageable amounts of data.
• Data capture, processing and storage: Depending on the
requirements of horizontal scalability and relational limitation
and the nature of their data, companies will generate a specific
IoT type that will require the selection of appropriate big data
technology for storage, access and processing.
• Data focus: The MTDC needs to focus on areas perceived to
have the greatest impact on its business. Again, the wide net that
can be cast using big data analysis technology means that the
perceptions can be tested to ensure they remain correct.
• Lack of capabilities and integration skills: The majority of MTDC
providers don’t have a team of experts on big data and analytics,
and during the initial phase they may rely on third parties to
overcome this skills shortage.
The learnings from IoT analytics will transform an MTDC provider,
from the management team to the data center infrastructure.
The process of becoming a data-driven organisation requires a
complete transformation that aects culture, capabilities, processes
and infrastructure. It is not limited to the data center even for a
company whose business is based on the data center.
The appetite for more knowledge about big data and analytics has
steadily increased year by year, as the emergence of thousands of
big-data education programs shows. As a result, many companies
have deployed big data analytics initiatives related to marketing,
finance and operations, and the number of use cases are growing.
Working in a broader big data analytic strategy that includes the
data center will become increasingly important for all companies
and act as a source of business for the MTDC provider.
Figure 12: Sensors monitor rack temperature distribution which
is then depicted as a real time heat map.
Source: Siemens, 2017