Trade-offs

Backs Act 3: the honest parts.

Questions to answer

• Dependency tax. How many services depend on netzlive.cluster, and what did the last major upgrade actually cost — PRs, deploys, calendar time?
• Distributed monolith. Where do other people draw the line? Collect sources, not opinions.
• Boilerplate. The full list of what a new service needs before its first line of business logic. Has anyone published a generator or starter kit for this?
• When not to cluster. Concrete cases at NETZlive where HTTP or Kafka would have been the better call.
• Alternatives. Partisan, Horde, :pg, and when each changes the picture.