If I am wrong in my understanding, please feel free to correct me

In my project, we have a timeseries database. It is setup as 3-node (One leader, 2 read-replicas) patroni cluster. Each node is an AWS EC2 instance where time-series data stored in hypertables supported by TimescaleDB extension on a Postgres database. We are using opensource timescaledb here.

As the data is growing each day, the EBS data volume on a node (EC2 instance) is expected to hit its size limit in future. Hence there need of sharding.

As the potential solution, we looked at distributed hypertables in timescaledb. But it seems to be a dead end as they have deprecated multi-node support (on top of which distributed hypertables are provided) in opensource timescaledb.

There is another option i.e. to use Citus (Postgres extension to implement sharding). But, Citus doesn’t support TimescaleDB extension in Postgres. So, as high level solution in this case, we have to convert timescaleDB hypertables to regular Postgres tables first for us to able to use Citus. So far Citus seems to be most suitable (relatively) choice to implement sharding.

Could someone please suggest a better way?