Comments by "" (@ContinuousDelivery) on "5 DevOps Best Practices" video.

  1. I think it depends on what you need to know to qualify for release. The most definitive measurement is "everything works together because I tested it together". If the system is decomposed to the point where each component is "independently deployable" (Microservice) then fine, but I have to do work to defend that "independent deploy-ability". If I test my components independently, in their own pipeline, but also test them again in some "integration test" with other components, then this is just another monolith, except now I have constrained my optimisation choices. I could have parallelized the build in lots of different dimensions if it were a "single repo monolith" but if I separate each component, I no longer have those options. So Service-based pipelines, followed by detailed integration testing, seems like: 1) a monolith and 2) a sub-optimal monolith to me.
    5
  2. No there are more, but thanks for asking and stay tuned for a future episode when I will describe 6 more recommendations. All of the recommended practices are based on the correlative model for software dev team performance that is at the heart of the Accelerate book: "Accelerate, The science of Lean Software and DevOps", by Nicole Fosgren, Jez Humble & Gene Kim ➡️ https://amzn.to/2YYf5Z8
    2
  3. I think that your last point is a choice, but a good one. You can have good architecture, and relax the inbuilt protections if the team is very good and very disciplined. Risky, but certainly possible - A kind of "Running with scissors" model. I think that the link between Bounded-Context/DDD and Microservice is often missed. When you get it right, the conversations between the services, though technically encoded, make sense at the level of the biz problem - which IMO is a VERY nice property to have.
    2
  4. YT doesn't let me move it either, No worries :)
    2
  5. Thanks for the suggestion
    2
  6. Thanks, there is some interesting stuff being written on applying these ideas to ML.
    1
  7.  @randomclimber  This is a decent start I think https://martinfowler.com/articles/cd4ml.html I also like the work that I have seen from Emily Gorcenski https://www.youtube.com/watch?v=bFW5mZmj0nQ
    1