Original Title: "Turbine: Block Propagation on Solana"
Original Author: Ryan Chern
Original Translation: Sharon, BlockBeats
Author Introduction: The author of this article, Ryan Chern, was a former engineering intern at Solana and co-founder of Quantum Labs. He currently works at Blackhawk Research, a cryptocurrency research center.
Data availability is crucial for blockchain, as it ensures that nodes can easily access all necessary information for verification, thus maintaining the integrity and security of the network. However, ensuring data availability while maintaining high performance is a significant challenge, especially as the network scales.
Solana has met this challenge with its unique architecture design, promoting the continuous creation and propagation of blocks. This has been achieved through several key innovations, such as leader selection, Gulf Stream (eliminating the need for a memory pool), and Turbine (block propagation mechanism).
The continuity of Solana requires an efficient system to ensure that all validators receive the latest state in a timely manner. One simple approach is to have leaders directly transmit all blocks to every other validator. However, given Solana's high throughput, this method would significantly increase bandwidth and other resource requirements, while compromising decentralization.
Bandwidth is a scarce resource, and Turbine is Solana's clever solution for optimizing the dissemination of information from a given block leader to the rest of the network. Turbine is specifically designed to alleviate the pressure of exporting data from the leader to the network.
In this article, we will delve into the working principle of Turbine and its key role in the wider transaction inclusivity field of Solana. We will also compare Turbine with other data availability solutions and discuss open research avenues in this area.
Turbine is a multi-layer block propagation mechanism used in the Solana cluster to broadcast ledger entries to all nodes. The core idea behind Turbine has been discussed in academia for many years, and this was demonstrated in a 2004 paper published and recent work.
Unlike traditional blockchains, which send blocks in order or flood them to all nodes, Turbine uses a more structured approach to minimize communication overhead and reduce the load on each node. At a high level, Turbine breaks a block into smaller chunks and propagates them through a node hierarchy. Here, a single node doesn't need to communicate with all other nodes, but only with a few selected nodes.
With the expansion of network scale, this becomes increasingly important, as traditional propagation methods become difficult to maintain due to the necessary communication volume. Therefore, Turbine ensures fast and efficient data propagation in Solana. The speed of block propagation and validation is crucial for maintaining Solana's high throughput and network security.
In addition, Turbine solves the problem of data availability, ensuring that all nodes can access the necessary data to effectively verify transactions. This is achieved without requiring a large amount of bandwidth, which is a common bottleneck in other blockchain networks.
Turbine greatly improves Solana's ability to handle high transaction volumes and maintain a streamlined and efficient network structure by alleviating bandwidth bottlenecks and ensuring fast block propagation. This innovative protocol is one of the cornerstones of Solana's promise of a fast, secure, and scalable network.
Now, let's take a deeper look at the mechanism of Turbine and how it spreads blocks on the Solana network.
Before propagating the block (i.e. transmitting it to other validators in the network), the leader constructs and sorts the block based on the incoming transaction flow. Once the block is constructed, it can be sent to the rest of the network via Turbine. This process is called block propagation.
Then, the voting messages are passed between validators, and these messages are encapsulated in block data to meet the commitment status of "confirmed" or "finally determined". Confirmed blocks are blocks that have received the majority of ledger votes, while final blocks are confirmed blocks and have built more than 31 confirmed blocks on top of the target block. Here is a detailed explanation of the differences in commitment status. This part of the consensus will be discussed in future articles.

Turbine's visualization of its position in the Solana transaction lifecycle
When leaders construct and propose the entire block, actual data will be sent as fragments (partial blocks) to other validators in the network. Fragments are the atomic units sent between validators.
At a higher level, Turbine acquires fragments and sends them to a set of predetermined validators, who then forward these fragments to a new set of validators. The following diagram summarizes the continuous process of fragment propagation:

As shown in the above figure, although it is called "block propagation", the data is propagated at the fragment level.
In this example, Validator 1 is the designated slot leader. During its slot period (Validator 1 is designated as the leader for 4 consecutive slots), Validator 1 constructs and proposes a block. Validator 1 first breaks the block down into sub-blocks called fragments through a process called sharding.
Smash the block data into fragments of maximum transmission unit (MTU) size (the maximum amount of data that can be sent from one node to another without dividing it into smaller units), and generate corresponding recovery fragment encoding schemes through Reed-Solomon erasure. This scheme helps with data recovery and ensures the integrity of data during transmission, which is crucial for maintaining network security and reliability.
This decomposition and dissemination process ensures that block data is distributed quickly and efficiently in Solana, thereby maintaining high throughput and network security.
Before propagating fragments through the Turbine tree, they are first encoded using Reed-Solomon erasure coding, a polynomial-based error detection and correction scheme. Erasure coding serves as a data protection method that can recover original data even if some parts are lost or damaged during transmission. Reed-Solomon erasure coding is a specific type of forward error correction (FEC) algorithm.
Due to the fact that Turbine fundamentally relies on a series of downstream validators for packet retransmission, these validators may be malicious (by selecting incorrect data for replay to counter Byzantine nodes or receiving incomplete data due to network packet loss). Due to the retransmission tree structure of Turbine, packet loss within any network scope is even more complex, and the probability of packets failing to reach their destination increases with each hop.
At a higher level, if the leader transmits 33% of the data packets in a block as erasure codes, the network can discard any 33% of the data packets without losing the block. The leader can dynamically adjust this number (FEC rate) based on the network conditions and variables such as recently observed packet loss and tree depth within the network.
For simplicity, let's examine the fragment group with a FEC ratio of 4:4.

Fragmentation group example: 4 out of 8 data packets may have been tampered with or lost, but it will not affect the original data.
Data fragments are partial blocks of the original blocks constructed by the leader, while recovery fragments are Reed-Solomon generated erasure coding blocks.
Blocks on Solana typically utilize a 32:32 FEC (32 packets out of 64 may be lost without requiring retransmission). As described in the Solana documentation, the following are conservative network assumptions:
· Packet loss rate 15%
· 50k TPS generates 6400 fragments per second.
32:32's FEC rate generates approximately 99% block success rate. In addition, if leaders choose to increase the probability of block success, they have the right to increase the FEC rate.
Turbine currently uses UDP for block propagation, providing significant latency advantages. According to the validator operator, it takes 100 milliseconds to transfer 6 MB+ erasure-coded data from us-east-1 to eu-north-1 using UDP, while TCP takes 900 milliseconds.
Turbine Tree is a structured network topology used by Solana to facilitate efficient propagation of shards (encoded block data) between validators. Once shards are correctly encoded into their respective shard groups, they can be propagated through the Turbine Tree to notify other validators in the network of the latest state.

Each fragment group is sent via network packets to a special root node, which manages which validators are part of the first layer (beyond 1 hop). The following steps are then executed:
1. List creation: The root node aggregates all active validators into a list, and then sorts them based on each validator's stake in the network. Validators with higher stake weights will receive faster fragments first, allowing them to respond more quickly with their voting messages to achieve consensus.
2. List reorganization: Then the list is reorganized in a deterministic way. This will use a seed derived from the slot leader ID, slot, fragment index, and fragment type to generate a "Turbine tree" from the validator node set of each fragment. A new tree is generated for each fragment group at runtime to mitigate potential security risks associated with static tree structures.
3. Layer formation: Then, starting from the top of the list, the nodes are divided into layers. The division is based on the value of DATA_PLANE_FANOUT, which determines the width and depth of the turbine tree. This value affects the speed at which fragments propagate in the network. The current DATA_PLANE_FANOUT is 6.
As we all know, the Turbine tree ensures that each validator knows exactly where they are responsible for relaying the fragment. Assuming the current DATA_PLANE_FANOUT value is 6, the Turbine tree is typically a 4 or 5 hop tree (depending on the number of active validators).
In addition, if a node does not obtain enough fragments or the loss rate exceeds the FEC rate, the node can fallback to idle talk and perform repairs. Under the current implementation, nodes that lack sufficient fragments to rebuild blocks will send retransmission requests to the leader. In a deterministic turbine, any node that receives a complete block can send requests for the repair fragments needed by the requesting node, thereby further pushing data transmission down into the area of the tree that requests data.
The block propagation on Solana is different from Ethereum. Here are some high-level differences:
1. Solana's ideal bandwidth requirement (>1 Gbps) is significantly higher than Ethereum's (geth recommends >25 Mbps). This higher bandwidth requirement is attributed to Solana's larger block size and faster block time. Solana's design can effectively utilize the entire bandwidth to accelerate data transmission and reduce latency. Although the bandwidth peak can reach 1 Gbps, it is not consistently used at 1 Gbps. Solana's architecture is specifically designed to meet the high demand for bandwidth peaks.
2. Solana uses Turbine for block data propagation, while Ethereum uses the standard gossip protocol. On Ethereum, block data propagation is done in a simple way: each node communicates with every other full node in the network. Once there is a new block, the client verifies it by sending it to its peers and approving the transactions in that block. This mechanism is suitable for Ethereum because compared to Solana, its block size is smaller and block time is longer. When it comes to Ethereum L2 aggregation data (excluding validium), propagation also follows the gossip protocol, and block data is stored in the "calldata" field of the Ethereum L1 block.
3. Ethereum uses TCP (via the DevP2P protocol) for block propagation, while Solana uses UDP (with some community support transitioning to QUIC). There are some trade-offs to consider between UDP and QUIC:
· Compared with QUIC, the one-way nature of UDP leads to lower latency, so QUIC streams are needed. Discussions on implementing one-way streams into QUIC are ongoing;
· Advocates of QUIC claim that while custom control flow can be achieved through UDP, it requires a significant amount of engineering work, whereas QUIC alleviates this workload by natively supporting such functionality. The ultimate goal is the same, but the performance ceiling of QUIC (latency, throughput, etc.) exceeds that of pure UDP in its current state.
These differences highlight the unique architectural decisions adopted by Solana and Ethereum, which help to improve their respective performance, scalability, and network robustness. For a deeper analysis of TCP, UDP, and QUIC, please see our article on Solana and QUIC here.
1. Discussions about Turbine as a "Data Availability" (DA) mechanism have surfaced. Turbine acts as a data availability mechanism, with the entire block data being published and downloaded by all other validators on Solana. However, Turbine lacks support for Data Availability Sampling (DAS), a feature that can help light nodes verify state while reducing hardware requirements. This is an active development focus for teams such as Celestia. Like Turbine, DAS also uses erasure coding, but the explicit purpose is to detect and prevent data withholding attacks.
2. For Solana Virtual Machine (SVM) L2s such as Eclipse, Tubrine has lost relevance as there are no validators set up to pass data between them. As for Eclipse, block data is published to Celestia for data availability - this allows external observers to run fraud proofs to ensure correct execution and state transitions. Eclipse will be one of the first implementations of SVM outside of the Solana network itself. Pyth also forked SVM for its oracle network "Pythnet" and runs it effectively as its own sidechain.
3. On Solana, full nodes manage block propagation and also participate in other parts of the integrated blockchain stack, such as transaction ordering and consensus. If Turbine runs as a modular component on dedicated hardware, what are its quantitative metrics?
4. Turbine prioritizes higher-weighted nodes to receive block data first. Over time, does this lead to more centralization of MEV?
5. In terms of raw throughput and minimal trust, how do different data availability methods (such as EigenDA (horizontally scalable single unicast relay) and Celestia (data availability sampling)) compare to Turbine in production?
Congratulations! In this article, we have studied Turbine and its workings in the broader transactional inclusivity field of Solana. We have compared Turbine with other data availability solutions and discussed different research approaches open to this field. Solana's Turbine protocol has demonstrated the network's commitment to achieving high throughput and low latency by effectively propagating block data between validators using a structured network topology.
The methods of enhancing data availability and improving block propagation efficiency can drive innovation within the wider blockchain community. A comparative analysis of the block propagation mechanisms of Solana and Ethereum reveals their respective advantages and trade-offs, and inspires deeper discussions on how emerging blockchain solutions such as EigenDA, Celestia, and Firedancer will shape this ecosystem in the future.
The solution for efficient data dissemination and data availability is far from complete. However, Solana's methods for optimizing network performance and its steadfast commitment have been warmly welcomed without compromising security and minimizing trust.
Welcome to join the official BlockBeats community:
Telegram Subscription Group: https://t.me/theblockbeats
Telegram Discussion Group: https://t.me/BlockBeats_App
Official Twitter Account: https://twitter.com/BlockBeatsAsia