Our archival software vendor (AAS) have a new version of their product. It replaces the performance bottleneck of Postgres with Cassandra. At the same time this opens up more possibilities for key/value storage next to the data, perhaps for extracted data.
I will provision some VMs for them to test with. Then we will soon upgrade our entire install, converting the storage to this new version. (New product name is Indigo)

The new version also has more flexible authN which is more suitable for API access by our testbed.