Hi,

if you use horizontal scaling with DDBs (which became the default quite some time ago), new DDBs are created to increase the amount of block references, a DDB can maintain.

if you reach the thresholds, otherwise the DDB will fail / go offline.

another option can be to add more partitions to the DDBs, increasing the number of unique blocks that can be handled.

but in the end, you always require SSD class disk space to store the DDBs ;)

There has been a registry key in the past, infuencing, whether a DDB is created as horzontal scaling or not. Since horizontal scaling gives you more backend capacity to address, I would not use that (any longer)
today you get DDBs for database, virtual server and file workloads, since the unique block matches between these workloads rarely match. 
This is resulting in maximum unique blocks per workload. 
additional DDBs will also increase the amount of unique blocks, that can be managed within a DDB engine, but lead to little higher backend consumption, due to redundant hashes possible in multiple DDBs belonging to the same engine. 
additional DDB partitions on the other hand, will not register unique hashes multiple times