# Verkle Trees - Preimages generation & distribution
---
## Why is this relevant for the Verkle conversion?
- We have to migrate all the data from MPT to VKT
- The keys in MPT are hashed
- We need the preimage of the hash to rekey it into the VKT
- Thus, **we need all the preimages of the MPT tree**
---
## Overlay Tree overview & timeline mental model
![image](https://hackmd.io/_uploads/rJCub9z7A.png)
---
## Current proposed strategy
![image](https://hackmd.io/_uploads/ryDMA5Mm0.png)
---
## How to decide the right moment for Slot C?
---
## Database generation and validation
- Who generates the database? (TBD)
- Can people easily verify it? (Yes)
---
## Database format
- Can be easy and efficient
- Flat file with [20 byte | 32 byte] ordered values
- No lookups, sequential read in the exact expected order
- Redundancy?
---
## How is the database distributed?
- Out of protocol:
- Multiple independent CDNs
- Torrent
- In protocol:
- Build sub-protocol allowing clients to share file through peers
---
## Most important question
- **How we distribute the preimage database? (In-protocol or out-protocol?)**
{"title":"Preimages slides","contributors":"[{\"id\":\"46f4f11e-c5fd-4f0b-9417-1132b8336f80\",\"add\":2161,\"del\":1002}]"}