Probably that logging in snip-snap, one-server-at-a-time, and click-clicking on various boxes in the (admittedly convenient) GUI leads to a bunch of unique snow

Probably that logging in snip-snap, one-server-at-a-time, and click-clicking on various boxes in the (admittedly convenient) GUI leads to a bunch of unique snowflake[0] servers that are prone to drift. If each server is doing one thing, that's obviously ok, but if you have 1000 web servers then treating them as immutable (aka phoenixes[1]) is conventional devops wisdom[2]. [0] https://www.ibm.com/developerworks/community/blogs/devops/entry/snowflakes_or_rubber_stamps?lang=en https://www.ibm.com/developerworks/community/blogs/devops/en... [1] https://www.thoughtworks.com/insights/blog/moving-to-phoenix-server-pattern-introduction https://www.thoughtworks.com/insights/blog/moving-to-phoenix... [2] https://news.ycombinator.com/item?id=4226099 https://news.ycombinator.com/item?id=4226099