Skip to main content
BitCurrent

Crypto markets, protocols and policy

Solana Alpenglow Goes Live, but 150ms Is Still a Target

Solana activated Alpenglow on mainnet, replacing TowerBFT with Votor and targeting 150-millisecond finality; outside measurements are slower so far.

BitCurrent Newsroom2 min read

Solana Alpenglow Goes Live, but 150ms Is Still a Target

Solana’s validators have used TowerBFT consensus since the network launched in 2020. The network activated Alpenglow on mainnet on Oct. 21, replacing TowerBFT with Votor, a new voting protocol designed to cut transaction finality to a fraction of a second.

  • Activation: epoch 1064, at slot 459,653,000
  • Protocol target: about 150 milliseconds, down from TowerBFT’s 12.8-second benchmark
  • Observed median: 404 milliseconds in OpenChainBench’s 24-hour measurement through Oct. 23

The Solana Foundation’s Alpenglow guide says validators began using the new consensus at the start of epoch 1064. It describes 150 milliseconds as a target, not a guaranteed result.

How does Votor change transaction finality?

Votor replaces on-chain validator vote transactions with direct votes between validators, which are bundled into certificates. A block becomes final after one or two rounds of voting, according to the Foundation.

That changes how quickly an application can treat a transaction as irreversible. Solana says transaction execution, programs and fees remain unchanged. The consensus upgrade affects agreement and finality, rather than how applications process transactions.

OpenChainBench measured a 404-millisecond median from slot arrival to root on a public RPC node over the 24 hours to Oct. 23. Its report says the network had been around seven seconds at the median under TowerBFT, making the observed improvement about 17 times.

Has Solana reached the 150ms target?

Not in that outside measurement. OpenChainBench says 150 milliseconds is Anza’s protocol target, while its own public-node reading was 404 milliseconds; the two figures measure finality from different points in the process.

The distinction matters because the target is not a report of typical user-facing performance. OpenChainBench measured the time from when its node first saw a slot until a root at or above it arrived.

Alpenglow’s rollout also has a second phase. The Foundation says Rotor, which is intended to replace Turbine’s block-propagation role, is planned for a later release but has no scheduled date. The mainnet activation and Votor handoff are confirmed; Rotor’s timing and the gap between the target and outside measurements remain open.

Source material