The Gluster has a couple dimensions of scale. When the service is scaled up the new containers are added to the pool, just not added to the volume. The created volume is just for data redundancy and keeps 3 replicas. Scaling it up by adding more nodes to the volume would only increase redundancy and not the storage capacity. I also think, though I am not a Gluster expert, it would degrade performance as the writes would need to replicate N times.
Once the service is operational, you could start managing Gluster as you would if it was not a containerized solution, and add additional volumes in different configurations. Essentially in Gluster you can create all the various RAID types. So you can get capacity/scale/redundancy.
The other considerations to make the current GlusterFS a scalable production ready solution would be to make use of IP reuse to allow it to more easily upgrade. Handle failure scenarios with additional health checks.
Hope this helps! We are continuing to work on it as well.