Anza, a blockchain infrastructure company oriented, has proposed an important overhaul for the Solana core protocol with the introduction of a new consensus mechanism that is dubbed.
According to a May 19 Blog postAnza described Alpenglow as the “biggest change” in the architecture of Solana since its foundation and called it a “turning point for Solana”.
The protocol would replace the existing TowerBft and proof-of-historical (POH) systems from Solana with a new architecture that is built around two important components, namely voters and rotor.
The voter handles the voice and block fininalization process, while Rotor serves as the data distribution protocol.
In contrast to the current design, which depends on Rossip -based communication and POH for time stamp, Alpenglow introduces direct messages and erases encrypted data planting.
Rotor builds on the existing turbine model of Solana, but reduces network shops and refines the selection of the Relais junction.
In the meantime, the voter introduces a double mode voting system that tries to complete blocks in a single round when at least 80% of the ring is active, or within two rounds when participation drops to 60%.
Both modes work at the same time, so that the protocol blocks can complete by which path first completes first.
Anza estimates that the upgrade can reduce median block finality to around 150 milliseconds, and as low as 100 milliseconds under optimum conditions.
These latency estimates, based on simulations, would bring the performance of Solana closer to those of the traditional Web2 infrastructure.
“A median latency of 150 ms does not only mean that Solana is fast it means that Solana can compete with web2 infrastructure in terms of responsiveness, making blockchain technology viable for completely new categories of applications that require real-time performance,” Anza’s Quentin Knentin, Kobi Sliwinshof.
The protocol also contains a “20+20” resilience model, which means that network safety and liveliness can maintain, even if up to 20% of the validators behave badly and another 20% react offline or not.
According to Anza, this design is responsible for both opponents and real circumstances, such as failure or latentary issues, aimed at consistent finality, even among degraded network performance.
Although Alpenclow is expected to introduce significant improvements in network reduction and resilience, Anza notes in his white paper That the upgrade alone will not prevent completely future disruptions, especially given the current dependence on the network on a single validator client.
Solana has struggled with network failure in recent years, often linked to congestion and validator over tax. The current architecture, which fogs a mempool and transactions directly, has made it particularly vulnerable to peaks in transaction volume, especially during periods of high activity.
Anza has played a key role in strengthening the infrastructure of Solana through his work on Agave, the primary validator customer of the network. The company continues to refine the performance of the customer, which most recently contributes to Agave V2.2, which increased the limits of the womb and improved transaction transport.
