We have used sharding for managing time series data e.g. 1 shard per day. Is there a way this could work i.e. where the number of shards continually grows?
However, CitusDB does support range partitioning and has a \stage command that will create new shards from incoming data. If you periodically load data that corresponds to a particular time range (hour, day, week), CitusDB can easily handle creation of additional shards during each load.