Overall, what we typically recommend is to launch a new instance and configure it as desired (eg, with tools, histories, and datasets of interest for the workshop). Then share that instance (using the CloudMan share-a-cluster feature
-> small green icon next to the cluster name on the main CloudMan console) and create a number of replicas of the shared instance for use during the workshop. For RNA-Seq workshops, for example, we generally stick with about 10 participants per instance. Here's a page with some empirical-based recommendations regarding choosing instance types: https://wiki.galaxyproject.org/CloudMan/AWS/CapacityPlanning
Finally, Dave Clements (CC'd) has a lot of hands on experience with setups for various workshops so he may be able to chime in and provide additional feedback.
Hope this helps and let us know if you have any more questions,