User Manual

> White Paper | Best Practices in Digital Transformation
19
u The National Institute of Standards and Technology (NIST)
suggests that companies consider two crucial data typologies when
defining the methodology for working IoT: non-relational data
representation, that is when data is semi-structured rather than fully
structured and tabular, and horizontal scalability which is valuable
when the amount of data coming in increases very quickly since
the data on one particular subject (a user or a device) is held in one
place rather than being spread across multiple structured tables and
therefore separated across servers. This allows for better control
over availability, speed and cost. However, this method makes it
dicult to enforce relationships between records and therefore to
preserve data integrity. Ultimately, therefore the decision needs to
be related back to corporate priorities for the data and the need to
preserve data relationships. While the choice of non-relational data
representation and scalability may be an obvious decision for high-
volume, simple and real-time analytics, in the situation within an
MTDC where dierent data sets (operational and CRM, for example)
need to be related to provide necessary value, then the decision
may not be so obvious.
This provides a comprehensive understanding of the dierent big
data scenarios that can be found in an organisation. Within an
organisation, dierent types of IoT data can be found, depending on
the business problem to be solved, which means that an organisation
might need to implement multiple big data technologies.
The need to capture, process, store and analyse data to generate
corporate value has generated the emergence of a new breed of
technologies. Critical to the process of using IoT are two main
categories of technology – for storage given the huge amount
of data involved, and processing. Under the first category come
NoSQL data stores, Apache Hadoop, Microsoft HDInsight, in-
memory databases and distributed file systems. Under the second
are Massive Parallel Processing (MPP), Hive, Big data in Excel.
These two sets of technologies work in pairs to store and process,
sometimes with a third access technology such as Polybase, Presto
and Sqoop.
IoT will play a critical component of more companies in future as
organisations seek to monitor and optimise the data lifecycle.
Ecient use of IoT in this context requires the data source to be
defined. All elements of the MTDC may be relevant data sources
to examine. Each of them provides valuable information for
understanding the performance of infrastructure and enables
infrastructure to be optimised. Ultimately, machine and software
data is the key to unlocking analytic applications. Analysis
conducted by DCD in 2016 [Cavanagh] indicates a number of areas
of infrastructure worthy of particular attention:
• Power – Elements of the power infrastructure are the electrical
service entrances of buildings, the main distribution unit,
generators, uninterruptible power supply (UPS) systems and
batteries, surge protection, transformers, distribution panels and
circuit breakers.
• Cooling – Systems that remove heat from the data center
include computerroom air-conditioning units (CRACs) and their
associated subsystems, chillers, cooling towers, condensers,
pump packages, piping, and rack- or row-level cooling or air-
distribution devices.
• Cabling – Data cables use dierent materials and connectors
to optimise performance and flexibility, while the ecient
management of the system maintains this optimisation for the
long haul.
• Racks and physical structure – The most critical of these
elements are the racks, which house IT equipment; physical-
room elements, such as dropped ceilings and raised floors; and
pathways to manage cabling.
• Facility management – Provide visibility of all physical
components. Management systems include building-
management systems, network management systems,
data center infrastructure management software and other
monitoring hardware and software.
• Grounding – This covers the common bonding network and
ground gear that protect equipment from electrostatic discharge.
• Security and fire protection – Subsystems included here are
physical security devices at the room and rack level and fire-
detection or suppression systems. u
Figure 10: A Model for the Deployment of IoT Technologies in MTDCs
Source: DCD 2017
Data Sourcing
Operational, Business,
Budgetary, CRM,
Machine data from IT
Systems, Infrastruture,
Storage, Networks
Data Center IoT data
storage, governance,
database access,
processing
Learnings
Customer development,
DT Services, DevOps,
Business Model
Outputs
IT Operations,
Networks, Infrastructure,
Monitoring, Alarms,
Capacity Planning
Analytics
Buisness analysis;
operational intelligence,
Customer intelligence