User Manual

> White Paper | Best Practices in Digital Transformation
18
The Internet of Things & the MTDC
“In a few decades’ time, computers will be interwoven into almost
every industrial product”. [Karl Steinbuch, 1966]
Dealing with IoT and capitalising on the opportunities it presents is
a core element of digital transformation. As a subset of ‘big data’, it
is far more important than humans as a source of data trac. It is
estimated that by 2020 there will be 50 Zettabytes of data generated
by more than 25 billion connected devices.
This represents an opportunity for MTDCs, since it oers insight that
can transform the way employees, suppliers, customers, products
and processes are understood and how facilities are designed and
operated. However, companies mention that they have struggled to
understand and communicate the benefits IoT can generate for their
data center and determine how to deploy a strategy. The perception
that digital transformation is unmanageably large and imposing can
be linked directly to perceptions and experience of IoT.
IoT is a key part of digital transformation. For the MTDC it will extend
the knowledge of and interaction with clients and lead to more
eective decision making in relation to them. It will also facilitate
that process in relation to external links into the data center such as
networks and other components of a client’s data infrastructure if
the MTDC is to act as the core control point of that infrastructure.
This role will take on an added importance if the MTDC is to operate
within a core-edge computing configuration.
The situations in which IoT can be of greatest value to the data
center will reflect the previously stated transformation strategy and
include operations, CRM, business development, infrastructure
utilization, virtualization, capacity planning, operating systems
and data products. Depending on the business opportunity to
be realised, an MTDC may need to implement a number of IoT
technologies and strategies.
All of the characteristics of the huge growth in IoT data will put
pressure on the processing, storage and analysis capabilities of the
MTDC:
• Big data - the phenomenon of which IoT is the major
contributor - is described in terms of:
• Volume refers to the amount of data generated. A decade
ago, data storage for analytics was counted in terabytes. Now,
organisations require at least petabytes of storage and the data
capacity of larger MTDCs may be measured in zettabytes.
• Velocity refers to both the throughput of data and its latency.
The legacy era was characterised by regularly batching data, now
access and processing occurs in real time measured in terms of
gigabytes or terabytes per second. Latency relates to the delay
between the data ingestion and the data analysis (measured in
terms of milliseconds).
• Variety refers to both the number of data sources and
the heterogeneity of data (structured, semi structured or
unstructured). Traditional analysis has dealt usually with
homogeneous data sources. Variability is a factor in this – this
is the phenomenon whereby data units can constantly change
meaning therefore requiring a variable analysis framework.
• Veracity means ensuring that the data is accurate and that data
that does not conform to the standards required of accuracy
does not accumulate in systems.
• Visualisation is important since a clear window on the data
makes it easier to validate and use for decision making.
• Value is the objective of the IoT analysis process and just as other
projects and processes are subject to ROI criteria so should this
process be.
Since the format and structure of IoT data can be established in
advance, the issues of variety, variability and veracity may have less
impact than for human sources of data. Nevertheless, IoT data from
a number of dierent sources may need consistent protocol and
language to be included within the same analytic processes. This is one
of the key challenges facing the development of edge processing.
The deployment of IoT will require:
• Sensors at the point where data is collected,
• Sensor technology to manage and direct the sensors,
• RFID tags,
• Embedded systems technology to direct the data flow,
• The means of analysis of the data through IoT analytics,
• The means of acting on the findings through human intervention
and/or AI, machine learning or cognitive solutions. u
Figure 9: The Classification and Analysis Requirements of Big Data
Non – relational data
representation PLUS
horizontal scalable
solution analysis
requirement
Non – relational data
representation analysis
requirement
Horizontal scalable
solution analysis
requirement
Velocity and/or Volume
Plus Variety
Velocity Only
Or Velocity
and Volume
Volume Only
Variety Only
Source: DCD 2017