I am benchmarking Sawtooth using Hyperledger Caliper. My network has 4 nodes and following configurations:
total transactions =4000
transactions/batch = 50
supply rate of transactions = 4000 transactions/sec
no. of clients = 1
type = smallbank
Consensus = PBFT
See the number of batches per block in this image: block list
Initially, the number of batches per block increases then decreases and then stays constant at 1 batch/block. This behavior is affecting my benchmarking results. Any idea why this is happening? I changed the block_publishing_delay to:
sawtooth.consensus.pbft.block_publishing_delay=0
, and sawtooth.publisher.max_batches_per_block=1200
.
How can I allow more batches per block? Please explain why there is only 1 batch/block for the later batches?