In traditional blockchains, block propagation occurs through a peer-to-peer network where new blocks and transactions are transmitted to all propagating nodes. This process is usually done sequentially or through mass broadcasting. However, this method is not scalable, because as the network grows, the amount of communication becomes difficult to manage.
Let’s consider a network of 35,000 validators. The leader needs to transmit a 128 MB block (approximately 500,000 transactions of 250 bytes each) to all these validators. The traditional block propagation implementation would require the leader to establish a unique connection with each validator and transmit the full block 35,000 times, resulting in a total data transmission of 4.69 TB. This exceeds typical bandwidth capabilities and becomes unfeasible with so many connections.
To solve the problem of slow block diffusion, Solana introduced Turbine, a multi-layer block propagation mechanism used by clusters to transmit ledger entries to all nodes. In general terms, Turbine divides blocks into smaller fragments and disseminates them through a specific order of nodes, reducing the load on each individual node.
Turbine was inspired by the BitTorrent architecture, as both use data fragmentation and peer-to-peer networks to distribute information. Turbine is optimized for real-time data transmission using UDP, which reduces latency. Additionally, it uses a random path per packet through the network while leaders transmit their data, allowing Solana to maintain high throughput in block propagation.
Furthermore, Turbine solves the data availability problem, ensuring that all nodes have access to the information needed to validate transactions efficiently, avoiding the bandwidth congestion typical of other blockchain networks.
Before propagating a block, the leader builds and organizes the block based on the incoming transaction flow. Once built, it is sent through Turbine to the entire network in a process known as block propagation. Messages are transmitted between validators and include the necessary information to reach a confirmed or finalized commitment state.
Although leaders create and propose entire blocks, the actual data is sent in partial fragments called shreds to other validators. Shreds are the smallest units of data transmitted between validators. Through this fragmentation and propagation process, fast and efficient distribution of the block across the Solana network is achieved, ensuring its high performance and security.
Before being sent to the Turbine structure, shreds are encoded using Reed-Solomon, a data protection method based on error-correcting codes. This technique allows information to be reconstructed even if certain fragments are lost during transmission.
Because retransmission of packets between validators could lead to the spread of incorrect or incomplete data, Turbine incorporates a dynamic strategy to compensate for packet loss. If a leader transmits 20% of the packets as error-correcting codes, the network can lose that percentage of data without affecting block reconstruction. This parameter can be adjusted according to network conditions.
Solana uses a tree structure known as the Turbine Tree to facilitate efficient propagation of shreds among validators. This hierarchical structure allows nodes to know exactly what their responsibility is in retransmitting fragments, ensuring optimized data distribution across the network.
If a node does not receive enough shreds due to loss exceeding the error correction limit, it can request retransmissions from the leader or from other nodes that have the complete block, ensuring continuity in the propagation of information.
In conclusion, Turbine enables efficient block distribution in Solana by dividing data into smaller fragments, facilitating its transmission without congesting the network. The implementation of error correction techniques and its tree structure ensure data integrity and availability.
At Q2BSTUDIO, we specialize in developing advanced technological solutions, including blockchain and distributed networks. Our team of experts works on optimizing digital infrastructures to ensure more efficient and scalable processes within the technological ecosystem.
