User Manual

> White Paper | Best Practices in Digital Transformation
21
Analytics
Analytics is the exhaustive use of data, statistical and quantitative
analysis, explanatory and predictive models, and data-driven
management to drive decisions and actions. Analytics can be
the input used for human decision-making or it may drive fully
automated decisions. In short, analytics enables decision-making
based on data and evidence, rather than speculation and ‘hunch’.
Analytics is important because data on its own is simply a set of
numbers (or other measurement units). Data itself needs to be
transformed to be able to be useful. The most usual spectrum to
describe this process moves from data through to wisdom as the
level of understanding progresses:
There are many benefits of deploying analytics for IoT within
the data center. These are linked with the reasons for deploying
IoT since analytics forms the processing component of the data
generated by IoT. The benefits can be described in terms of the
evolution of strategy through dierent types of analysis:
• Descriptive analysis – The MTDC provider is able to measure
what happened in the data center. For example, such analysis
can show how much electricity the data center has used over
the past week. This enables the provider to gain operational
visibility across data center infrastructure.
• Diagnostic analysis – The organisation is able to understand
why certain events happen in its data center – such as why the
data center consumed more power last week than this. Events
can then be linked to causes. Causality enables strategies to be
developed on the basis of being able to influence events.
• Predictive analysis – The organisation is able to predict certain
data center interactions – namely, how much data center energy
a particular client will need or what the impact will be of raising
the temperature at which the IT is run by 1
o
C. This requires more
complex analysis and usually some form of statistical modelling
to establish how monitoring infrastructure in real time and
correlating events across layers leads to certain outcomes, and
with what degree of certainty these outcomes happen. These
processes apply also to the next two forms of analysis.
• Prescriptive analysis – The organisation is able to make decisions
related to its data center based on scenarios. For example, it can
identify data-center energy optimisation strategies.
• Preventive analysis – The organisation is able to act in advance
of data center needs, such as increasing data center capabilities
based on a public cloud.
It is worth noting that MTDCs need a structured, actionable path
toward optimising their facilities in terms of business objectives by
leveraging data and analytics for their decision-making.
Developing a data center analytics strategy involves an evolution
through dierent stages that include description, diagnostics,
prediction, prescriptive analysis and prevention. In addition to
helping MTDCs to find solutions to practical operational problems,
analytics in the data center opens the door to new data-driven
opportunities in terms of CRM and service development.
The deployment of IoT analytics will entail a number of significant
process issues that need to be overcome including data generation,
data quality, data capture, processing and storage, IoT focus and
the availability of both required capabilities and integration skills.
The development of analytics needs to be based on a longer-term
strategic view and through a process which allows learnings to be
incorporated with flexibility into the direction. There needs also to
be a cultural change whereby intuition and hunch move towards
data-based decision making.
It will have also HR implications. MTDCs have, historically, put a
greater focus on the facility than on IT but their IT departments
will also need more employees, not fewer, and they will need to
look also at employing the new breed of experts in terms of Data
Science/Analytics, DevOps and CRM. These are skills sets now in
considerable demand. In HR terms, the process will erode the value
of the facility and IT silos, and require the redefinition of roles based
on workloads and objectives.
“Typically on-prem is reserved for the most sensitive of data – much
of our systems that we manage are client owned, we treat these
are part of our internal systems in terms of the way we govern
their operation – it is highly complex some on-prem is owned
and operated by us and then we have cloud services – some of
which are contractually owned by clients and some by us and
simply resold. Across each service model we have varied delivery
methods – we tend to have a dedicated IT team for each contract
and for each major location. We have a strengthening anity with
cloud delivery and seek to deliver cost savings not only for our own
Figure 11: The Data Assimilation Spectrum
Data Information Knowledge Wisdom
Relations Patterns
Principles
Source: DCD 2017
u