Ethereum vs Solana uptime.

The scoreboard

First, the numbers:

The rest of this article is why they look the way they do, and what each one is actually counting.

What a “halt” is, and why it’s the right metric

A halt is when a network stops producing blocks and cannot resume on its own. Recovery requires validators to coordinate out of band: they agree on a restart point, roll back to it, and bring the network up again together. Nothing settles while a chain is halted. No transactions, no trades, no liquidations.

This is the metric that matters, because it is the one thing a blockchain is supposed to guarantee. Fees, throughput and latency are quality-of-service questions. Whether the chain is producing blocks at all is the availability question, and that’s what this comparison is about.

Ethereum’s record

Zero halts since genesis on 30 July 2015. The network has produced blocks continuously through a $150M contract hack, a sustained denial-of-service campaign in 2016, the worst congestion in its history on Black Thursday, a client bug that split the chain in 2020, and a live switch from proof of work to proof of stake in 2022.

The single closest call came on 11–12 May 2023, when the beacon chain twice stopped finalising, for roughly 25 and 64 minutes. It kept producing blocks the whole time and recovered on its own, with no restart. That is the worst it has ever been, and it was not a halt. The full incident history is here.

Solana’s record

Solana is a genuinely fast chain, and for stretches an extremely popular one. It has also stopped. The widely reported network-wide outages, each of which required validators to coordinate a restart or left the chain effectively unusable for hours:

Counting is a little fuzzy at the edges. The January 2022 episode was prolonged degradation rather than one clean stop, which is why “around seven” is the honest number rather than an exact one. The shape is not fuzzy: repeated network-wide stops, several of which needed the validator set to restart the chain by hand.

Why the two records differ

The gap is not luck; it comes from deliberately different designs.

Client diversity vs a dominant client

Ethereum is run by many independent client implementations, written by different teams in different languages. A bug in one client is an inconvenience for the operators running it, not a network halt, because the rest of the network keeps producing blocks. Solana has historically leaned on a single dominant validator client, so a single bug can take down the whole network at once. It has, more than once. That mechanism has its own article.

Conservative throughput vs maximum throughput

Solana targets very high throughput and low fees, and pushes hardware and software hard to get there. That design is fast, and it has also historically been more brittle under load: several of its outages trace back to bursts of traffic the network could not absorb. Ethereum keeps base-layer throughput deliberately modest and pushes scale outward to layer-2 rollups, trading peak speed for a base layer that has never fallen over.

The fair version

Solana is not a bad network, and that is not the claim here. It is often cheaper and faster than Ethereum’s base layer, and its reliability has clearly improved as the software has matured. If your priority is raw throughput and low fees, that is a real and defensible trade. But on the narrow question this site cares about, whether the chain has ever stopped, the records are simply not comparable. One has never halted. The other has, repeatedly, and needed help to come back.

And Bitcoin?

Bitcoin has also never halted since its 2009 launch. On pure availability, Bitcoin and Ethereum sit in the same category. Solana is not there yet.