By default, there is three environment variables that can be run with ruby roxy configuration of a MarkLogic (the roxy script is here: https://github.com/marklogic/roxy). The three variables are: local, dev and prod. Although, others environment variables can be added.
What would be the best setup for local, dev and prod servers? The prod server is on an ec2 AWS.
I think, locally, there should only be 1 or 2 forests, and on a server maybe 1 or 2 forests for 4 cores. The forests can function in parallel. I think in the dev environment, tests should be run. Also, interested to know how much space the MarkLogic DB caches when run, both locally and on a ec2 instance.
I believe, by default the prod environment variable does run test (although that might be wrong).
Some information:
As mentioned by Dave, best setup depends on many factors, but I'd add that nr forests primarily depends on the amount of data you'd like to load, and also on the response times, and loading speeds you seek. With the same amount of data, more forests will allow more parallelization of querying and ingestion. Once you reach limits in cpu, mem or disk speed, you scale out horizontally to allow for additional forests.
Regarding, local versus dev and prod, that really depends on how much data you like to load on each, and the limits of the system itself. I typically start off with 3 forests per host (kind of a default with Roxy), even for local.
And yes, like Dave said too, I'd encourage reading Inside MarkLogic Server. It goes pretty deep, and has recently been updated to include latest features.
HTH!