Part of #33. Depends on the Distribution primitive.
Problem
Calldata size is fixed per scenario (0 / ~68B / ~2KB Disperse). No way to vary payload size or sweep size buckets.
Approach
- A
sizeDist (histogram/buckets) drives calldata pad length, independent of the key distribution — so a workload can sweep size while holding contention fixed, and vice versa.
- Wire into
StorageRW (e.g. writeN(key, pad)) or a padded-transfer scenario.
Acceptance criteria
- Per-tx payload size drawn from configurable weighted buckets.
- Size and key distributions are orthogonal (settable independently).
- Stresses surfaced (mempool/gossip bandwidth, state-write volume) distinct from execution contention.
Design: https://github.com/sei-protocol/platform/blob/main/docs/designs/sei-load-workload-modeler.md
Part of #33. Depends on the Distribution primitive.
Problem
Calldata size is fixed per scenario (0 / ~68B / ~2KB Disperse). No way to vary payload size or sweep size buckets.
Approach
sizeDist(histogram/buckets) drives calldata pad length, independent of the key distribution — so a workload can sweep size while holding contention fixed, and vice versa.StorageRW(e.g.writeN(key, pad)) or a padded-transfer scenario.Acceptance criteria
Design: https://github.com/sei-protocol/platform/blob/main/docs/designs/sei-load-workload-modeler.md