What Is Sharding? Blockchain Scalability Explained 2026
Key Insight
Sharding is a blockchain scaling technique that divides the network into smaller partitions (shards) that process transactions in parallel. Instead of every node processing every transaction, different shards handle different subsets. Ethereum shifted its sharding roadmap toward danksharding, which focuses on providing cheap data availability for Layer 2 rollups. Proto-danksharding (EIP-4844) introduced blob transactions that reduced L2 fees by 90%+.
Sharding is one of the most anticipated scaling solutions in blockchain technology. By dividing the network into parallel processing units, it promises to dramatically increase transaction throughput without sacrificing decentralization.
What Is Sharding?
Sharding is a scaling technique borrowed from database engineering that divides a blockchain network into smaller partitions called "shards." Each shard processes a subset of the network transactions independently and in parallel with other shards.
In traditional blockchains like Bitcoin and early Ethereum, every node must process every single transaction. This creates a bottleneck: the network can only be as fast as a single node. Sharding breaks this limitation by distributing the workload.
Related: What Is Ethereum?
Database Sharding Origins
Sharding originated in distributed databases:
The Restaurant Analogy
Imagine a restaurant with one cashier processing all orders:
- No sharding: One cashier handles every customer sequentially
- Sharded: Four cashiers each handle customers with different last name ranges (A-F, G-L, M-R, S-Z)
- Result: 4x more customers served in the same time
Database Implementation
Large databases like Google Spanner and Amazon DynamoDB use sharding:
- Split data across multiple servers
- Each server handles a portion of queries
- Coordinate across shards when needed
- Enables horizontal scaling
Types of Blockchain Sharding
Network Sharding
Split validators into groups:
- Each group validates different transactions
- Reduces communication overhead
- Maintains random assignment for security
- Validators rotate between shards
Transaction Sharding
Distribute transactions across shards:
- Transactions assigned to shards based on rules
- Each shard processes its subset independently
- Cross-shard transactions require coordination
- Increases overall throughput
State Sharding
Split blockchain state across shards:
| Approach | Description | Complexity |
|---|---|---|
| ---------- | ------------- | ----------- |
| Network sharding | Split validators | Low |
| Transaction sharding | Split transaction processing | Medium |
| State sharding | Split entire blockchain state | High |
State sharding is the most powerful but most complex approach, as different shards store different parts of the global state.
The Blockchain Scalability Trilemma
Sharding attempts to solve the scalability trilemma:
- Security: Network must resist attacks
- Decentralization: Many independent validators
- Scalability: High transaction throughput
Traditional blockchains sacrifice scalability for security and decentralization. Sharding aims to improve scalability without compromising the other two.
Security Challenges
With sharding, each shard has fewer validators:
- Easier to attack a single shard
- Random validator assignment mitigates this
- Frequent reshuffling prevents collusion
- Minimum shard size ensures security threshold
Ethereum Sharding Roadmap
Original Plan: Execution Sharding
Ethereum originally planned 64 execution shards:
- Each shard processes smart contracts independently
- Cross-shard communication via beacon chain
- Massive complexity in implementation
- Required solving cross-shard composability
The Pivot: Rollup-Centric Roadmap
Ethereum pivoted to a rollup-centric approach:
- Layer 2 rollups handle execution
- Base layer focuses on data availability
- Simpler implementation
- Better security model
Danksharding
Named after researcher Dankrad Feist:
- Single proposer selects data for all shards
- Data availability sampling (DAS) verifies data
- Rollups post transaction data cheaply
- No independent execution shards needed
Proto-Danksharding (EIP-4844)
What It Is
Proto-danksharding, implemented in Ethereum Dencun upgrade (March 2024), introduces "blob transactions":
- Blobs: Large data chunks attached to transactions
- Temporary storage: Blobs pruned after ~18 days
- Separate fee market: Blob fees independent of gas fees
- Rollup-optimized: Designed for L2 data posting
Impact on Layer 2 Fees
| Network | Pre-EIP-4844 | Post-EIP-4844 | Reduction |
|---|---|---|---|
| --------- | ------------- | --------------- | ----------- |
| Arbitrum | $0.50-2.00 | $0.01-0.05 | ~95% |
| Optimism | $0.50-2.00 | $0.01-0.05 | ~95% |
| Base | $0.30-1.00 | $0.005-0.02 | ~97% |
| zkSync | $0.40-1.50 | $0.01-0.04 | ~96% |
Proto-danksharding dramatically reduced the cost of using Layer 2 networks.
How Blobs Work
- Layer 2 rollup batches user transactions
- Creates a blob containing transaction data
- Submits blob transaction to Ethereum
- Blob stored temporarily (not in execution state)
- Validators verify blob availability
- After ~18 days, blob data pruned
- Commitment remains for verification
Related: What Are Gas Fees?
Full Danksharding
Data Availability Sampling (DAS)
The key innovation of full danksharding:
- Validators only sample random portions of blob data
- Statistical guarantees ensure data is available
- No node needs to download all data
- Enables massive data throughput
Target Specifications
- Blob count: 64-128 blobs per block (vs 6 in proto-danksharding)
- Data throughput: ~1.3 MB per block
- Effective L2 scaling: 100,000+ TPS across rollups
Implementation Timeline
Full danksharding requires:
- Data availability sampling protocol
- Proposer-builder separation (PBS)
- KZG commitment infrastructure
- Peer-to-peer data distribution
- Validator client updates
Sharding on Other Blockchains
Near Protocol
- Dynamic sharding based on demand
- Nightshade protocol
- Chunks processed in parallel
- Cross-shard transactions built-in
Elrond (MultiversX)
- Adaptive state sharding
- Three shard types (network, transaction, state)
- Secure proof of stake consensus
- Cross-shard execution in 2 blocks
Zilliqa
- Pioneer of blockchain sharding
- Network and transaction sharding
- Linear scalability with network size
- DS committee coordinates shards
| Blockchain | Sharding Type | Shards | TPS Achieved |
|---|---|---|---|
| ----------- | -------------- | -------- | ------------- |
| Ethereum | Data (danksharding) | Planned 64-128 | Via L2 rollups |
| Near | Dynamic state | Adjustable | ~100,000 |
| MultiversX | Adaptive state | 3+ | ~15,000 |
| Zilliqa | Network + transaction | Variable | ~2,800 |
Sharding vs Layer 2 Solutions
Complementary Approaches
Sharding and Layer 2 are not competing solutions:
- Sharding: Makes base layer better at data availability
- Layer 2: Handles execution off-chain with base layer security
- Combined: Sharded L1 + L2 rollups = maximum scalability
Why Ethereum Chose Both
Ethereum determined that:
- Pure execution sharding is extremely complex
- Rollups already solve execution scaling
- Data availability is the real bottleneck
- Danksharding provides data, rollups provide execution
- Combined approach is simpler and more effective
Challenges of Sharding
Cross-Shard Communication
Transactions spanning multiple shards require:
- Coordination protocols
- Atomic commit mechanisms
- Increased latency for cross-shard operations
- Complexity in smart contract composability
Data Availability
Ensuring all shard data is available:
- Validators must prove data exists
- Data withholding attacks possible
- DAS provides probabilistic guarantees
- Erasure coding adds redundancy
Implementation Complexity
- Years of research and development
- Multiple client implementations needed
- Extensive testing required
- Backward compatibility concerns
Key Takeaways
Sharding is a fundamental scaling technique that divides blockchain workload across parallel partitions. Ethereum evolved from execution sharding to data-focused danksharding that complements Layer 2 rollups. Proto-danksharding already reduced L2 fees by 90%+ and full danksharding will enable 100,000+ TPS across the rollup ecosystem. While complex to implement, sharding combined with rollups represents the most promising path to blockchain scalability without sacrificing security or decentralization.
Continue learning: What Is Ethereum? | What Are Gas Fees? | Complete Blockchain Guide
Last updated: February 2026
Sources: Ethereum.org, EIP-4844, Vitalik Buterin Blog
Key Takeaways
- Sharding splits blockchain data across parallel partitions for higher throughput
- Ethereum pivoted from execution sharding to data-focused danksharding
- Proto-danksharding (EIP-4844) reduced Layer 2 fees by over 90%
- Blob transactions provide temporary cheap data storage for rollups
- Full danksharding will scale Ethereum to 100,000+ TPS through L2s
Frequently Asked Questions
What is sharding in simple terms?
Sharding is like splitting a large database into smaller pieces so multiple processors can work on different pieces simultaneously. In blockchain, instead of every computer processing every transaction, different groups handle different transactions in parallel, dramatically increasing overall capacity.
How does sharding improve blockchain speed?
Without sharding, every node processes every transaction sequentially. With sharding, the network divides into groups that work in parallel. If you have 64 shards, theoretically you get 64x more throughput because 64 groups process transactions simultaneously.
What is the difference between sharding and Layer 2?
Sharding is a Layer 1 change that modifies the base blockchain itself. Layer 2 solutions build on top of the existing blockchain. Ethereum strategy combines both: danksharding makes the base layer better at storing L2 data, while L2 rollups handle actual transaction execution.
What is danksharding?
Danksharding is Ethereum specific sharding approach named after researcher Dankrad Feist. Instead of creating independent execution shards, it focuses on providing massive data availability for Layer 2 rollups. This approach is simpler and works better with the rollup-centric roadmap.
Has Ethereum implemented sharding yet?
Ethereum implemented proto-danksharding (EIP-4844) in the Dencun upgrade of March 2024. This introduced blob transactions that dramatically reduced L2 costs. Full danksharding with data availability sampling is planned for future upgrades and will provide even greater scale.
