Solana Franchit un Cap Historique : 100 000 Transactions par Seconde en Conditions Réelles

Le réseau Solana vient de pulvériser les records de scalabilité. Lors d'un test grandeur nature sur son mainnet, la blockchain a atteint 100 000 transactions par seconde - du jamais vu dans l'écosystème crypto.

Performance brute contre réalité décentralisée

Ces chiffres astronomiques font pâlir les solutions traditionnelles comme Visa (24 000 TPS théoriques). Mais les puristes rappellent qu'il faut distinguer la performance en environnement contrôlé des contraintes du monde réel.

La scalabilité extrême de Solana repose sur son mécanisme de proof-of-history combiné au proof-of-stake. Un pari technologique qui commence à porter ses fruits, malgré quelques blackouts mémorables par le passé.

Pendant ce temps, les banques centrales peinent toujours à faire tourner leurs CBDC à 50 transactions par seconde... Ironie du sort pour ceux qui voyaient dans la blockchain un gadget inutile.

Solana tests transaction completion limit

The noop feature enables solana to fill the blocks with light instructions that burn out the network capacity. 

Developers use this approach to test the throughput limits without experiencing failures in complex applications. Although these transactions demonstrate the upper end of Solana, they do not accurately represent use cases like payments, decentralized finance (DeFi) activity, or application interactions.

Mumtaz said that although the dataset was mostly composed of no-operation calls, the findings suggest that the network could potentially support 80,000 and 100,000 TPS in transfers, oracle updates, and other practical operations.

Solana’s current throughput data presents a much less impressive picture. Solscan estimates approximately 3,700 TPS in the blockchain, two-thirds caused by validator voting transactions. Validators submit votes multiple times per slot to maintain consensus, inflating reported throughput metrics.

The difference between stress-test levels and real-world averages marks the challenge between the potential theoretical top performance and maintainable actual performance. Although the 107,000 TPS is impressive and shows the efficiency of the Solana design, its actual throughput on transactions remains an order of magnitude lower.

The stress test is a continuation of the momentum of Firedancer, a validator client by Jump Crypto. Firedancer has testnet benchmarks exceeding 1.2 million TPS, but the client has not yet been released to mainnet. 

Other structural improvements being developed by engineers include decoupling consensus and execution and using localized fee markets to eliminate network congestion. Bitwise researchers highlighted the stability of Solana under prior stress tests, noting that the chain has the potential for high-frequency applications.

Community vote on Alpenglow consensus

Although the stress test highlighted network capacity, Solana validators are voting on a significant protocol change. The Alpenglow proposal (SIMD-0326) reached its community voting period on epoch 840, with ballots closing at the end of epoch 842.

The upgrade aims to replace the existing TowerBFT protocol with a high-performance and stronger consensus system. Alpenglow proposes a new variant of block propagation called Rotor to improve block finalisation time as well as decrease latency. Should it be adopted, Rotor would reduce confirmation lapse by 100-150 milliseconds once 80% of the validators are available.

Validators will be required to pay a Validator Admission Ticket (VAT) of 1.6 sol per epoch to keep the economy disciplined. This fee would be burned instead of redistributed, and voting does not increase the token supply. In every round, leaders would continue receiving rewards by reporting combined vote data and finalization certificates.

Get $50 free to trade crypto when you sign up to Bybit now