On Wednesday (21), Solana activated Alpenglow on its mainnet, replacing TowerBFT, the consensus mechanism used since the blockchain's launch in 2020. The upgrade changes communication between validators and seeks to reduce the time required to make a transaction final from approximately 12.8 seconds, in the historical reference presented by the network, to 150 milliseconds.
The Solana Foundation confirmed that the deployment occurred in epoch 1064, which began at around 13:36 UTC. The effective replacement of the protocol happened approximately 18 minutes later, at 13:54 UTC, when Alpenglow took over consensus at slot 459,653,000.
The transition was also documented by Unchained, which examined public blockchain data. The last block associated with TowerBFT was recorded at slot 459,652,999, at 13:54 UTC, and served as the reference for Alpenglow's initial certificate.
Next, eight slots passed without block production, until slot 459,653,008 was recorded, about three seconds later. The analysis identified continuity in block numbering, with no evidence of chain reversion during the migration.
The official documentation adds that the reference block received votes from 446 of the 654 validators counted on the network. These records confirm the transition was executed, but do not, in isolation, demonstrate that the new speed target is continuously maintained.
Votor is already active, but Rotor remains for a second phase
The component implemented in this first phase is Votor, responsible for voting and finalizing blocks.
Under TowerBFT, validators recorded votes through transactions on the blockchain. With Alpenglow, they now exchange signed messages directly with each other, gathered into certificates that allow blocks to be finalized after one or two rounds of voting.
The change eliminates the old vote transactions from blocks and alters the way infrastructure services monitor the network. The confirmation states confirmed and finalized now represent the same condition, while transactions-per-second metrics need to be recalibrated so that the removal of votes is not interpreted as a drop in user activity.
Validator operators also begin paying a fixed fee of 0.8 SOL per epoch to participate in the process, replacing the previous costs associated with individual voting transactions.
The second phase calls for Rotor, a protocol intended to replace Turbine in distributing block data among network participants. This stage does not yet have a defined deployment date.
Despite the structural change, the smart contract execution mechanism and the fees for ordinary transactions remain unchanged.
The Solana Foundation emphasizes that 150 milliseconds represent a finality target, not a guaranteed fixed time for all transactions. The result depends on factors such as communication between validators and the number of rounds required to reach consensus.
The comparison with 12.8 seconds also requires context: that value corresponds to the previous model with 400-millisecond slots, later reduced to 200 milliseconds in another upgrade.
Thus, the activation confirms the replacement of the consensus protocol, while Alpenglow's sustained performance on the mainnet still needs to be assessed by subsequent measurements.
