Industrial Internet: Towards Interoperability and Composability
Figure 1: Main approaches for achieving interoperability
The broker approach, akin to using an interpreter between two speakers of different languages,
works reasonably well with a small number of interacting parties (after tolerating the potential
problem of ‘lost in translation’). It is perhaps the only way to enable communication between
parties ‘after the fact’ – that they were previously built with different specifications and now find
the need to communicate. This is a task we are facing in connecting many legacy industrial
systems that have previously been deployed. The broker approach clearly has its shortcomings
in scalability both in term of the number of different ‘species’ of interacting parties requiring
specific brokering (design time complexity) and the sheer number of parties in a deployment
requiring brokering (runtime complexity). It also suffers in stability because a broker would
require updating whenever any of the ‘species’ changes or new ‘species’ are added to the mix.
The common meta-model approach removes the scalability problem found in the broker
approach but requires foresight – interoperability by design – as the interacting parties must be
created with the common meta-model and agreed-upon interfaces.
There are a number of ways of achieving common meta-models and agreed interfaces. The
prevailing approaches include:
1. Common Specification: The interacting systems are built from common specifications as
a consensus of technical communities, often in the form of Open Standards available to
all implementers free of charge or at a nominal fee. The modern communication systems
serving as the foundation of the Internet (e.g., RFCs from IETF, specifications from IEEE
and ETSI, etc.) and the application stacks in the IT systems (e.g., the SQL specification from
- 62 -
June 2016