Using Metrics in the Industrial IoT Value Chain to Drive Trustworthiness
� Understand the nature of shortcomings and failures of a system component so that these can be mitigated within the system, or negotiated with the provider.
Figure 1: The Value Chain of Industrial IoT
understanding the key considerations in a system and also aid in the analysis and design of a system, especially if historical metrics data is used to inform decisions. Clearly metrics can also be used during the operation of a system to maintain visibility into its operation and to help ensure that trustworthiness and other business and operational targets are met.
Good metrics will typically find many uses. For example, metrics on storage service reliability will help to:
� |
Clarify the expected service level and |
quality with any service providers. |
Metrics |
also |
support |
contractual |
enforcement such as assessment of |
penalties in case of failure to fulfill SLA |
( Service Level Agreement) or SLO |
( Service Level Obligation) targets, and |
support |
the |
ability |
to |
compare |
|
providers. |
|
|
|
|
� |
Evaluate |
how |
well |
the |
service |
performs. This in turn allows for |
precise feedback to providers. |
A real world example of this third point is a system where edge devices periodically invoke the cloud storage service directly every minute, while having the capacity to handle a backlog of only up to ten minutes of data stream. In this scenario, it is important to prevent downtimes of nine minutes or more of the data storage service. An adequate metric measuring the duration of downtimes – not just the uptime average – will be the basis for negotiating service quality in order to minimize data loss.
This paper describes how operational metrics data may be combined with business and risk management information to support a better understanding of trustworthiness, enabling investments made in trustworthiness to be better managed.
MANAGING TRUSTWORTHINESS
Trustworthiness in the business context means that a satisfactory level of confidence can be established in any system component( be that a sensor, a machine or a factory). Confidence can be established in what it claims to be, whether it fulfills its ascribed tasks, has appropriate performance, and will not endanger people, the environment, partners or the organization due to any issues relating to security, safety, reliability, resilience and privacy.
Tradeoffs and decisions need to be made among business and functional requirements as well as design decisions and
September 2018- 89-