Event-Driven APIs on specific platforms

This chapter adds requirements and best practices for specific SBB platforms

Table of contents

  1. Requirements Kafka Platform
    1. MUST comply with the Naming Conventions for Topics
  2. Best practices Kafka Platform
    1. Initialload / state restore

Requirements Kafka Platform

MUST comply with the Naming Conventions for Topics

Infrastructure artifacts like topics and queues must be named according to the following naming conventions:

{application-abbreviation}.{application-specific}

  • {application-abbreviation}:
    [a-z0-9-]+ (Sequence of:lower case,numbers, dashes). Important: For internal applications, use one of the documented names or aliases according to the EA Repository.
  • {application-specific}:
    [a-z0-9-.] (Sequence of:lower case, numbers, dashes, periods)

Best practices Kafka Platform

Initialload / state restore

When building an AsyncAPI on the Kafka platform, you should ask yourself how a prospective customer can restore its state, using only the Kafka topics as a source. After all, the feature of sending and storing messages in a longterm fashion is the unique selling point of this technology. Consider the following hints

  • Are you sending self-containing snapshots or events (which require some kind of a baseline)?
  • Are there additional topics where your customer may obtain a baseline in a periodic fashion?
  • Are the retention time on the topic appropriate for this use case?

If possible, refrain from building a backchannel to request a baseline. There is also always the possibility to build a baseline topic with Kafka Streams