It sacrifices decentralization and it has poor reliability.

The post you’re replying to discusses and measures centralisation.

I’m happy to discuss reliability but it doesn’t appear we’re having a conversation.

I have no horse in this game but the Wikipedia page [1] doesn’t line up with the story you’re painting:

> On 30 April and 1 May, 2022, Solana experienced major outages due to being taken offline by bots.[18] On 1 June, 2022 Solana reportedly[19] suffered another outage due to a bug in how the blockchain processes offline transactions. On 1 October, 2022, the Solana network went down for 6 hours due to a single misconfigured node.

In a properly decentralized system, how does 1 misconfigured node take down the entire system? That looks like a single point of failure to me, something decentralized systems aren’t supposed to have. Can you elaborate how your distinguishing reliability and centralization in the context of single node failures?

[1]: https://en.m.wikipedia.org/wiki/Solana_(blockchain_platform)

Sure - it’s a bug, it’s been fixed. The issue occurred on 30 Sep and there’s a discussion here:

https://solana.com/news/09-30-22-solana-mainnet-beta-outage-...

That’s how software works. Things break. They particularly break at scale. Nobody’s done this before. If Eth was as busy it would break too.

They have done it before -- Bitcoin existed first, and has kept clearing transactions every ten minutes since, and will continue to do so once Solana is long dead.

Bitcoin and Eth have nowhere near the usage of Solana.

and Solana has way less usage than the average postgres DB running webapp xyz

Nope: https://news.ycombinator.com/item?id=33879077

I don't see how your link disputes my claim

ok

Ethereum and Bitcoin don't break because they thought about "What happens if we get DDOS'd" and added a fee market, something Solana still hasn't properly addressed which is why it still falls over when attacked by spam.

A network falling over because of high usage or (in this case) one validator had a misconfigured node is extremely bad.

Eth and Bitcoin don't have high usage. And the issue with the single validator mentioned, as you'll be aware from reading https://github.com/solana-labs/solana/pull/28172, wasn't a design failure, but a software issue. Bitcoin and Eth have those too.

There's a lot to like about Solana, but one thing to dislike is the ongoing claim by the Solana community that Solana is actually fairly decentralized.

In reality, Solana is a highly centralized blockchain. Solana has zero protocol specs (the code is the spec), one production client, one main development organization (Solana Labs), limited exposure to the academic research community, and the entire network only runs in data centers due to extreme hardware & bandwidth requirements.

If we wanted to boil down decentralization into a narrow metric (we shouldn't want this, but if we did), that metric might be "how many guns to how many heads to force an invalid state transition?" And Solana scores very poorly on this metric.

These are good things to be concerned about. There's a second client and a second validator in the works, there's some stats on datacenter decentralisation in the youtube link above https://www.youtube.com/watch?v=5NDs3Il2J3Q&t=303s - 230 datacenters, 9% in a single data center. But yes lots of work to be done - I don't want to pretend there isn't - but also don't want to pretend that a network with far less activity has has already done it.

I may be wrong today, but I took a peek awhile back and Solana's top N validators -- enough to exert outright control over the network -- were operated by companies from within the same 2-3 VC portfolios, which struck me as a shockingly poor standard of "decentralization" for a network which apparently prides itself on that metric.

From a certain angle, it could almost look as if said VCs were trying to add legitimacy to their portfolio investments by colluding to make Solana seem busier and safer than it really is...