In an Archimate diagram, would Kafka be an application component or a system software according to the specification? The literature gives contradictory information:

• Kafka as system software: for example: “In the Archimate, Kafka
  would be categorized as “system software”. System software represents
  software that provides or contributes to an environment for storing,
  executing, and using software or data deployed within it. It’s a
  specialization of a node that models the software environment in
  which artifacts run”.

• Kafka as application component: most authors consider Kafka as
  application component, for example: “In the context of ArchiMate,
  Kafka would typically be considered an “application component”. An
  application component represents an encapsulation of application
  functionality aligned to implementation structure, which is modular
  and replaceable. It can include not only reusable software components
  but also complete software applications, sub-applications, or
  information systems. So, Kafka, being a self-contained unit that
  encapsulates behavior and data, aligns with the definition of an
  application component. Its behavior would be modeled explicitly
  through relationships to behavior elements.”

• Kafka as application component and system component: some
  authors recommend using an application component and relate Kafka as system
  software with Realization-relationship to it.

• When I look at examples for system software Archimate specification, it mentions “OS, JEE Application Servers, DBMS, Workflow Engines, and
  Communication Middleware”. For my understanding, Kafka has very much
  in common with Communication Middleware (producers write into topics, consumers write from them) and DBMS (especially if used with fact events). Then it would by a system software.

What is the best choice and why?