For low-latency use-cases and for easier implementation of non-deletion proofs, the Unicity aggregator should support a sharding mode where each shard communicates directly with the BFT cluster and the BFT validators merge the shard sub-trees to obtain the partition root hash. There is currently no requirement for combining this sharding mode with the one implemented in unicitynetwork/aggregator-go#54 (to deploy a sharded aggregator, the operator must choose one of the two sharding modes).
For low-latency use-cases and for easier implementation of non-deletion proofs, the Unicity aggregator should support a sharding mode where each shard communicates directly with the BFT cluster and the BFT validators merge the shard sub-trees to obtain the partition root hash. There is currently no requirement for combining this sharding mode with the one implemented in unicitynetwork/aggregator-go#54 (to deploy a sharded aggregator, the operator must choose one of the two sharding modes).