If you are familiar with how blockchains operate, then you know that they are generally made up of nodes that store the entire state of the currency’s history and verify all transactions.
Flow, by contrast, aims to create many subdivisions of its network to allow for the total work to be split across nodes, with each node only validating a subset of the transactions.
Architecture
To do this, the Flow blockchain employs a multi-node, multi-role architecture.
Put another way, Flow has divided the validation stages of a transaction into four different categories, dividing the responsibilities of each node:
Collection Nodes – Enhance network connectivity and data availability for dapps
Consensus Nodes – Decide the presence and order of transactions
Execution Nodes – Perform computation associated with each transaction. These nodes do not have any decision making powers
Verification Nodes – Double check the work done by the Execution Nodes.
Flow’s creators argue this level of specialization allows each node to participate in the validation of every transaction, while splitting tasks to increase efficiency.