Introduction to Ethereum's Data Architecture
In Ethereum's ecosystem, data is ultimately stored as [key,value] pairs using LevelDB as the underlying database. The blockchain organizes transactions into blocks, which link together to form BlockChain and HeaderChain structures. At a granular level, transactions and contracts operate within these blocks, while account states are managed through stateObjects and StateDB.
Core Data Units
- [k,v] Database: LevelDB serves as the foundational storage layer
- Blocks: Contain transaction lists and execution results
- BlockChain/HeaderChain: Linked block structures at different abstraction levels
- stateObjects: Individual account representations
- StateDB: Manages the collection of all accounts
1. Block Structure: Header vs. Body
Block Header Components
The header contains critical metadata about the block:
| Field | Description |
|---|---|
ParentHash | Pointer to parent block |
Coinbase | Miner's address |
Root | State trie root hash |
TxHash | Transaction trie root hash |
ReceiptHash | Receipt trie root hash |
Difficulty | Mining difficulty value |
Number | Block height |
👉 Explore Ethereum block structure
Block Body Components
The body contains the actual transactional data:
type Body struct {
Transactions []*Transaction
Uncles []*Header
}Key features:
- Transactions: List of all executed transactions
- Uncles: Stale block headers included for network stability
2. Merkle-Patricia Trie (MPT): Ethereum's Data Organizer
MPT Node Types
| Node Type | Description | Use Case |
|---|---|---|
fullNode | Branch node with 17 children | Hex path navigation |
shortNode | Compacted path node | Patricia trie optimization |
valueNode | Leaf node (32-byte hash) | Final data storage |
hashNode | Node reference (32-byte hash) | Merkle proof linking |
MPT Operations
func (t *Trie) Insert(key []byte, value []byte) {
// Recursively navigate and modify trie
}
func (t *Trie) Hash() common.Hash {
// Compute root hash via recursive hashing
}3. Storage Architecture
Layered Data Management
StateDB: Business-layer cache
- Manages stateObjects in memory
- Batches writes to trie
MPT Trie: Intermediate cache
- Organizes accounts by address
- Commits to database periodically
LevelDB: Persistent storage
- Final [k,v] storage layer
- Optimized for fast lookups
👉 Understanding Ethereum storage
Version Control System
type revision struct {
id int
journalIndex int
}
func (s *StateDB) Snapshot() int {
// Create restore point
}4. HeaderChain vs. BlockChain
Structural Comparison
| Feature | HeaderChain | BlockChain |
|---|---|---|
| Contents | Headers only | Full blocks |
| Memory Use | Lightweight | Heavy |
| Validation | Fast sync possible | Full validation |
| Use Case | Light clients | Full nodes |
FAQ: Ethereum Data Management
Q: How does Ethereum ensure data integrity?
A: Through MPT's cryptographic hashes - any change alters root hash
Q: Why separate Header and Body?
A: Headers enable lightweight verification while bodies contain full transactional data
Q: What's the purpose of Uncles in blocks?
A: They improve network decentralization by rewarding stale blocks
Q: How are state changes committed?
A: Through StateDB's intermediateRoot() and CommitTo() calls
Q: What makes MPT unique?
A: Combines Patricia trie's path compression with Merkle tree's cryptographic verification
Conclusion
Ethereum's data architecture demonstrates sophisticated engineering balancing:
- Cryptographic security via Merkle proofs
- Storage efficiency through Patricia tries
- Performance optimizations via layered caching