So the new datastream concepts are slightly confusing. I'd like to play a bit with the details and possible configurations. Can I create my own datastream type? For example some documentations say only logs and metric are allowed (https://github.com/elastic/logstash/issues/12178).
So is it allowed to create custom datastream types? I'm thinking for example of a flow datastream type, which could hold sflow and netflow data.
As Filebeat does not support sflow (please add support), we have to use the Logstash sflow input codec (Elastiflow). Can we use Logstash to create datastreams or is that not supported yet?
Another question, the documentation is talking about auto-generated backing indices. So how does this mix exactly with the normal non-hidden indices and what's the purpose of the metrics-index_pattern_placeholder and logs-index_pattern_placeholder placeholder indices?
My current experience with hidden indices is not very good, considering some of the newly created hidden indices are generated on warm nodes instead of on hot nodes, as wildcard patterns don't match on hidden indices...
We currently are not using Agent yet. Can we use datastreams with Beats or is Agent a requirement?