| cip | 34 |
|---|---|
| title | Set Pruning Window to 7 Days + 1 Hour |
| description | Sets the minimum data pruning window for DA bridge nodes to 7 days + 1 hour |
| author | Callum Waters (@cmwaters), Nashqueue (@Nashqueue) |
| discussions-to | https://forum.celestia.org/t/cip-lower-data-pruning-window-to-7-days-1-hour/1965 |
| status | Final |
| type | Standards Track |
| category | Data Availability |
| created | 2025-03-17 |
| requires | CIP-036 |
| supercedes | CIP-4 |
Abstract
This CIP proposes to set the minimum MinimumPruningWindow for Celestia nodes to 7 days + 1 hour (169 hours). This change is intended to complement CIP-036, which is expected to define a 7-day SamplingWindow. A pruning window of 7 days + 1 hour ensures that nodes can serve data for the entire sampling period. This adjustment aims to reduce storage requirements for node operators while maintaining data availability for light clients.
Parameters
| Parameter | Current value | Proposed value | Description |
|---|---|---|---|
| MinimumPruningWindow | 721 hours | 169 hours | The minimum retention period of block data for serving the network (nodes can set higher values) |
Rationale
The next pending increase in throughput (32 MB blocks every 6 seconds) would require bridge nodes to store up to 1 TB per day. The current pruning window (30 days) implies that pruned bridge nodes need to have 30 TB of storage. This CIP reduces the pruning window to 7 days + 1 hour to reduce the storage requirement to 7 TB which makes it less costly to serve data to the network.
This CIP focuses solely on the MinimumPruningWindow. The rationale and specification for the 7-day SamplingWindow itself are detailed in CIP-036.
Backwards Compatibility
This CIP’s proposed change to the MinimumPruningWindow should be implemented after the changes specified in CIP-036 (which defines the 7-day SamplingWindow) are adopted by the network. Nodes must first support the new SamplingWindow before the MinimumPruningWindow is reduced to 7 days + 1 hour to ensure light nodes don’t sample correctly pruned data.
Copyright
Copyright and related rights waived via CC0.