NEAR’s blockchain uses a new consensus algorithm and a sharding architecture to achieve a relatively high level of performance. The key elements of NEAR’s technology are:

• Dynamic Re-Sharding
• Nightshade
• Doomslug
• Fees as Developer Rewards
• Token Balance-Based Storage

Dynamic Re-Sharding There are two ways to approach scalability - Horizontal and Vertical. NEAR has chosen to use sharding, which is a type of horizontal scaling technique. Sharding is an approach to scaling databases, and it involves spreading computing and storage capacity across multiple servers. In the world of blockchains, sharding involves partitioning a blockchain across subgroups of nodes and distributes computation across multiple parallelized sub-chains ("shards"). Sharding is not live on NEAR at the moment.

Once sharding goes live on NEAR, the network intends to regularly adjust the number of shards supported by the network based on user demand. The project calls this demand-based approach to scaling "Dynamic re-sharding." This scheme allows the network to only pay for the infrastructure and scale it needs at any given time. Dynamic re-sharding might be a more cost-efficient approach to scaling and securing a sharded network since node storage requirements can adjust with demand.

Nightshade NEAR uses a new sharding approach called Nightshade to reach consensus and settle cross-shard transactions. Most sharded networks use a smaller validator committee that rotates its lead block proposer when validating transactions across shards. This approach introduces some difficulties to the validation process. For instance, these validator groups must either download the entire state of each shard or receive only the portion that changed. The validation process for either technique can be inefficient as the first one requires a bigger time frame and the second one requires a larger state size.

Nightshade offers a different model by modifying the typical sharding abstraction. It assumes that each shard works together to produce "chunks" of data that collectively create a single block. Blocks are produced with a regular cadence regardless of whether each individual shard has produced its “chunk” for that specific block height. Nightshade is a leader-based system, meaning it assigns a single validator to produce each block. This validator must assemble the data chunks generated by each shard during their leadership period into a block. Leadership assignments will rotate through the existing validator set. Validators don't accept transactions, only chunks.

Doomslug Doomslug is NEAR Protocol's new block production technique. According to the team, it enables the network to achieve some sense of practical finality after one round of communication, with a finality gadget providing BFT finality after a secondround. Practical finality in Doomslug (or Doomslug finality) is when a block is irreversible unless at least one participant is slashed. Doomslug also allows the network to continue producing and finalizing blocks as long as over half of the validator set are online and honest, although the finality gadget will stall if less than two-thirds of participants are online.

Fees as Developer Rewards NEAR distributes 30% of any transaction fee to the contract (or contracts) that transaction calls. The contract’s owner (often a developer group or a DAO) can decide how to allocate these funds. These fee “rewards'' effectively incentivize early application development on NEAR. The percentage of fees allocated to this reward is a system-level parameter (can be adjusted by governance), but developers may choose to receive any amount equal to or above this minimum value.

Token-Based Storage NEAR token holders have the right to store data on the chain. For example, if someone holds a balance of 1 NEAR, it means that the person can store nearly 10 KB of data on the account. This model is similar to how checking accounts in banks require a minimum balance to have an account. It allows contracts that are maintaining important state to pay validators proportionally to the amount of data they are securing.