Sharding

Sharding is a scaling technique that splits a blockchain's workload, data or state into smaller partitions called shards. Different groups of nodes handle each shard in parallel, so no single node has to process everything.

The idea comes from database sharding, where a large database is split across many servers. In a sharded blockchain, validators are divided into groups, and each group handles the transactions and state of one shard. A coordinating layer ties the shards together so they share the network's security. Because no single node needs to process every transaction, total capacity can grow as shards are added.

For people using crypto, the goal is less congestion and lower fees as more users and applications join a network. NEAR Protocol is one example of a blockchain designed around sharding. Ethereum took a different path. Its earlier plans for many separate shard chains gave way to a rollup-centric roadmap. EIP-4844, known as Proto-Danksharding, added blobs, which give rollups dedicated space to post their data. Full Danksharding aims to expand that data space much further.

A common misconception is that Ethereum is being split into many chains. Neither Danksharding nor Proto-Danksharding follows the traditional model of dividing the blockchain into parts. They focus on data availability for rollups instead. Sharding in general also has real trade-offs. Transactions that move between shards are complex to coordinate and can be slower. Spreading validators across shards also means each shard is guarded by fewer of them, which could make a single shard easier to attack. Designs counter this by assigning validators to shards randomly and rotating them.

Learn more

Related terms