So if using queue mode, you need redis, am not sure if they support others, but redis is the middle man in syncing with workers and main n8n, btw enterprise has ha such as multi main, but scaling wise queuemode etc.
So u can have multi workers which checks the redis queue that main posts there, note the data in redis is very small just references to workflow to run from db etc etc,
Also if using filesystem mode helps alot more than running all in RAM ie for long workflow it can eat up ram quickly causing ooms, crashes. typically it just stops the execution for that wf but not always.
And you can have few other options too. But hope that helps, btw, you can always build ontop of n8n infra, and inside a workflow use MQs incase of crashes etc, so u can send crud redis queue or rabbit kafka etc