Qaynaq LABS · TOOLS

CRYPTO GLOSSARY

Sharding

A scalability technique that splits a blockchain into smaller, parallel chains called shards, each processing a portion of transactions to improve throughput without compromising security.

Technical definition

Sharding is a database partitioning strategy adapted for blockchains, where the network is divided into smaller subsets called shards. Each shard contains its own subset of nodes, transaction history, and state, allowing parallel transaction processing. Validators are assigned to specific shards, and cross-shard communication protocols (e.g., receipts or merkle proofs) enable transactions involving multiple shards. Security is maintained through random validator assignment and cryptographic proofs to prevent collusion or single-shard takeover attacks. The goal is to increase transaction throughput (transactions per second) while preserving decentralization and security, though it introduces complexity in consensus, data availability, and cross-shard coordination.

An analogy

Imagine a library with thousands of books, but only one librarian. If someone requests a book, the librarian must search every shelf sequentially, causing long wait times. Sharding is like hiring multiple librarians, each responsible for a specific section of the library. Now, multiple book requests can be handled simultaneously, reducing wait times. However, if a book spans multiple sections, the librarians must coordinate to retrieve it, adding a bit of complexity.

Why it matters

Blockchains face a scalability trilemma: they can only optimize two of decentralization, security, and scalability at once. Sharding aims to improve scalability (throughput) without sacrificing the other two, addressing congestion and high fees on networks like Ethereum. It enables broader adoption by supporting more users and applications without requiring centralization.

Where it came from

The concept of sharding in blockchains was popularized by Ethereum’s research team around 2018–2019 as part of the 'Ethereum 2.0' (now Ethereum 2.0 or 'The Merge') upgrade. Earlier, sharding was explored in traditional databases (e.g., Google Spanner) and other cryptocurrencies like Zilliqa, which launched the first sharded blockchain in 2019.

Example in practice

Ethereum’s 'Beacon Chain' introduced sharding in its 'Dencun' upgrade (2024), where the network is split into 64 shards. Each shard processes transactions independently, and data from shards is periodically committed to the main chain (Beacon Chain) for finality. This allows Ethereum to handle ~100,000 transactions per second (theoretical) compared to ~15–30 pre-Merge.

Common misconceptions

  • Sharding reduces security → Correction: Sharding can maintain security if designed properly with random validator assignment, cryptographic proofs, and economic incentives to prevent attacks like single-shard takeovers.
  • Sharding makes blockchains fully centralized → Correction: Sharding distributes work across many nodes, but poor design (e.g., few validators per shard) could centralize control. Proper sharding requires a large, diverse validator set.
  • Sharding is only for Ethereum → Correction: Sharding is a general technique used by other blockchains like Zilliqa, Near Protocol, and Polkadot (which uses a similar but distinct 'parachain' model).

Risks & limitations

  • Single-shard takeover attacks: If validators controlling a shard collude, they could manipulate transactions or censor users in that shard.
  • Cross-shard communication failures: Complex protocols for transactions spanning multiple shards may introduce delays, bugs, or security vulnerabilities.
  • Data availability issues: If shards do not publish data correctly, users may be unable to prove transactions occurred, leading to disputes or lost funds.
  • Validator centralization: If sharding requires large staking amounts, it could reduce the number of validators, increasing centralization risk.
  • Network fragmentation: Poorly designed sharding may create isolated shards with incompatible states, reducing interoperability.

Related terms

About "Sharding" in crypto

Cryptocurrency vocabulary moves faster than any other technical field. New terms appear weekly, old terms quietly change meaning, and most "definitions" online were written by someone trying to sell you something. Qaynaq Labs' crypto glossary aims for the opposite: neutral, accurate, skeptical explainers that include the downside.

This entry covers the technical mechanics, a real-world analogy, where the concept came from, and what could go wrong. Pair it with the related terms to build a working mental model of how the pieces fit together.

Frequently asked

Does sharding make a blockchain as secure as a non-sharded one?

Not inherently. Sharding can maintain security if designed with strong cryptographic guarantees, random validator assignment, and economic incentives (e.g., slashing for misbehavior). However, poorly implemented sharding may reduce security by creating smaller, more vulnerable subsets of the network.

How does sharding compare to Layer 2 solutions like rollups?

Sharding and Layer 2 solutions address scalability but at different layers. Sharding improves base-layer throughput by parallelizing transaction processing, while Layer 2 solutions (e.g., Optimistic or ZK rollups) process transactions off-chain and post compressed data to the base layer. They can complement each other: sharding increases base-layer capacity, while rollups handle complex computations.

Can sharding be used for privacy-focused blockchains?

Yes, but with additional complexity. Privacy-focused blockchains (e.g., Monero, Zcash) could theoretically use sharding to scale, but privacy features (e.g., confidential transactions) make sharding harder to implement. Solutions like 'sharded ZK-SNARKs' or 'private sharding' are still experimental.

What happens if a shard fails or goes offline?

A failed shard cannot process transactions until it recovers. Depending on the design, the network may pause transactions involving that shard or route them to other shards. Recovery typically requires the shard to resync with the network, which may take time.

Does sharding require changing the consensus mechanism?

Not necessarily, but it often does. Sharding is often paired with a consensus mechanism that supports parallel processing, such as Proof of Stake (PoS) or a modified version of Proof of Work (PoW). For example, Ethereum’s sharding uses PoS for validator assignment across shards.