Architecture
The Tezos network is composed of many Tezos nodes running different jobs with the overall goal of maintaining the state of the data in its blockchain in a decentralized manner.
This diagram shows a high-level view of the Tezos system:
Components
These are the main components of Tezos:
- The data of the chain itself
- The nodes that store the chain data, accept transactions to add to the chain, and verify transactions that other nodes add to the chain
- The clients that send transactions to the nodes
The blockchain data
Although people often use the word "blockchain" to mean the entire system, strictly speaking, a Tezos blockchain is a series of blocks of data, each connected to a previous block in the chain. Tezos has a primary blockchain, referred to as Mainnet, and a group of blockchains used for testing, referred to as testnets. Anyone can create a new blockchain in the Tezos system, so nodes reach consensus on which blockchain is the primary one and which block should be the next block in any specific blockchain.
Technically speaking, there can be many slightly different copies of each blockchain. When one node adds a block, its copy of the blockchain becomes slightly different from that of other nodes that haven't heard about the new block yet. Also, a node might add an invalid block that other nodes later challenge and refuse to add to their copies of the blockchain. In this way, the blockchain is technically a tree of blocks, but the nodes constantly work to achieve long-term consensus on a straight line of blocks where each block has exactly one preceding and following block.
The data inside a Tezos block includes operations such as transferring tez between accounts, deploying or calling smart contracts, and adding messages to the smart rollup inbox. Blocks also include operations that are necessary for the management of the chain, including nodes' attestations that blocks are valid. For more information on the operations that can be in blocks, see Blocks and operations in the Octez documentation.
Nodes
Tezos nodes are computers that store and manage the state of a blockchain.
The main types of Tezos nodes are:
-
Baking nodes, which bake new blocks and send them to other nodes
In addition to baking new blocks, baking nodes are responsible for storing a copy of the current state of the blockchain data, known as the context. When they create a block or receive a new block from another node, they run each operation in the block, such as tez transfers and calls to smart contracts, and compute the new state of the context.
Baking nodes are also responsible for communication with other nodes via a peer-to-peer network. Nodes communicate over this P2P network, also called the gossip network, to exchange new blocks and operations.
-
Accuser nodes, which verify the blocks that other nodes bake and challenge blocks that have problems
The accuser monitors new blocks and looks for problems, such as when bakers try to add more than one block at a time. When it finds a problem, it submits a denunciation to other nodes to refuse the new blocks and punish the offending node.
-
Smart Rollup nodes, which run logic on a layer that is separate from the main Tezos baking process
Anyone can run a node and choose which node functions to run. Because different programs control functions such as baking blocks and challenging other nodes' blocks, your nodes can choose which programs to run.
Clients
Many clients can add operations to the network of nodes, including:
- Web applications
- Command-line applications such as the Octez client
- Wallet applications
- Applications that use SDKs such as Taquito
When a client adds an operation, that new operation becomes part of the mempool, which is a list of available operations. When nodes bake blocks, they take operations from the mempool and put them into a block. Nodes share operations with each other, so the node that bakes an operation into a block may not be the node that the client originally sent the operation to.
Layers
Tezos runs transactions on two layers:
-
Baking nodes run transactions on Tezos layer 1, which is the primary layer and the layer that most people mean when they talk about Tezos and transactions.
-
Smart Rollup nodes run transactions and logic in a separate environment that doesn't need to follow all of the same rules as layer 1. The transactions and logic that Smart Rollups run is called layer 2.
For example, Smart Rollups can run transactions at a much higher rate than layer 1 does. Also, they can run large amounts of processing and handle large amounts of data that would be impractical on layer 1. Smart Rollups regularly publish their state to layer 1 for nodes to verify.
Separating logic into layers like this allows Tezos to scale to run large amounts of transactions. Smart Rollups can receive messages from layer 1 and use the information in the messages to run large processing operations independently of layer 1.
For more information about Smart Rollups, see Smart Optimistic Rollups and the tutorial Deploy a smart rollup.
Communication between layers
As shown in the diagram above, each layer 1 block has a global rollups inbox that contains messages from layer 1 to all rollups. Anyone can add a message to this inbox and all messages are available to all rollups. Rollups receive this inbox, filter it to the messages that they are interested in, and act on them accordingly.
Smart Rollups can send messages back to layer 1 by adding them to their outbox, which consists of calls to smart contracts on layer 1. Unlike baking nodes, Smart Rollup nodes do not communicate with each other directly.
Smart Rollups also have access to the reveal data channel, which lets them request data from sources outside the inbox. This channel lets rollups access large amounts of data without having to store it on layer 1.
References
For more information about the architecture of Tezos, see: