What number of are “a number of”? My expertise in a considerably related atmosphere (*) together with just a few dozen groups and plenty of experiments working in parallel is that there is no such thing as a easy technical answer and it’s not at all times potential to know what induced a regression or enchancment.
What you are able to do is have a central, simply accessible and simple to know, board exhibiting what deployments are working in the meanwhile and what they embody, and educate folks to verify it earlier than they deploy.
You will get bonus factors by:
-
Be capable to lock elements of the system, both in a comfortable (e.g. the above board) or arduous means
-
Have a coordinator position, this could be a full time job or a rotating position. The later has the advantage of rising the shared duty feeling. Alternatively this could be a assembly or perhaps a mailing record the place modifications are printed
-
Have a “deployment schedule” per workforce or space
Word that a lot of the options would possibly result in decrease deployment frequency and challenges in implementing an actual CI/CD
(*) A/B testing in manufacturing of a posh system, not micro companies or clusters however every module was developed individually whereas retaining an agreed API. The ultimate pre-release testing was achieved utilizing one other A/B check group of testers.
