Ask HN: Is there any consensus on defining microservice size?
Put another way, how have you seen service boundaries defined, and what was the cost/benefit?Did you take a way a preference or set of criteria that were particularly effective?
From the linked article linked about 'Fast':
Earlier this year, NPR reported on how CEO Holland had his share of controversy in Australia prior to starting Fast. Holland’s former startup, Tow.com.au, which aimed to be “the Uber of towing,” failed in what at least one person described as a “disaster.” NPR’s article noted that Holland’s previous venture was embroiled “in a multimillion-dollar billing dispute with the Australian state government over towing and impounding fees that led to the startup’s liquidation in 2018.”
I highly recommend "Poverty, By America" if you'd like the background for my comment.
https://www.amazon.com/Poverty-America-Matthew-Desmond/dp/05...
Put another way, how have you seen service boundaries defined, and what was the cost/benefit?Did you take a way a preference or set of criteria that were particularly effective?
- Battle tested choices for libraries
- Documentation outside of our own (O'reilly books, online courses, etc..)
- SO posts about issues and sticky points
- Patterns for implementation (lack of these is shitty for JR devs. The only place to learn is PR feedback)
- Popular style guides
- IDE plugins
...and many more. Like being able to list POPULAR_LIBRARY_FROM_JOB_DESC on a resume when I can finally leave this job where leadership prioritized the browser's needs over dev comfort and productivity
Also, the advantages of web components and shadow DOM have never shown up. They only make CSS and browser testing more difficult