Hacker News Reader

Comment by britneybitch | Hacker News Reader

It feels like Solana is moving fast and breaking things a little too much for a project that has to do with money. Uptime should be a competitive advantage for decentralized networks, but the Solana network went down 12 times in the first 6 months of this year[1]. Even Rust couldn't save them.

[1]: https://www.datacenterdynamics.com/en/news/solana-cryptocurr...

View on HN

> Even Rust couldn't save them.

Rust doesn't stop bugs, it just helps you try reason about stuff and maybe catch them earlier.

To be fair to the Solana team, their production chain is called Mainnet Beta.

Going forward their outages should occur less frequently since they've finally introduced a fee market akin to Ethereum's dynamic gas prices. With a dynamic fee market it will be more financially prohibitive to DDOS the chain.

Did it go down 12 times for technical reasons or was it intentionally pulled down? Something to wonder about.

Is there a difference? A decentralized protocol that can be DOS'd by a reasonably successful criminal gang isn't all that useful.

It goes down for an incredibly stupid decision. Transactions on Solana are free so spammers can produce millions of transactions per second and at some point validators simply crash.

Solana has subsecond transaction processing so a validator being down for more than a second counts as a downtime whereas with Bitcoin a block taking two hours to mine is completely normal. Bitcoin doesn't have the concept of "downtime" only the concept of delaying the next block.

Doesn't sound very decentralized at all...

No it just had bugs that got fixed. Decentralisation does not imply perfection.

My understanding is that the outages are not bugs but a deliberate tradeoff between throughput and liveness.

No, they’re bugs, but Solana (average 4500 transactions per second) competes more with Visa (average 7500 transactions per second) and older blockchains aren’t even in the arena. The bugs get fixed, Solana’s pretty stable, you can see the exact numbers at https://status.solana.com/

Solana's definition of transaction is weird though. There's no way a P2P network is doing 4500 txs in the common blockchain terminology. The entire system is a house of cards waiting to crumble - just napkin the average tx size * 4500 txs * peers etc...

How are there enough transactions on Solana to get to 4500 per second? How is there possibly enough demand on a crypto network to be remotely in the same league as visa? I don't buy it.

A lot of Solanas throughput is consensus messages between validators.

But finance in general is many orders of magnitude more TPS than visa. Think about high frequency trading of derivatives of all sorts of things, and all the tiny arbitrages that can be done over all the exchanges when fees are cents. Those can easily multiply to thousands of TPS.

Wash trading bots can create as much volume as you want.

That’s literally what Sol was designed to do. Plus taking 20% of all NFT traffic in a year is a huge part of it. Even with consensus messages removed Sol still does more transactions than all the L2s combined.

But why is this the case when their volume is nowhere near that percentage of the L2 chains too? What exactly does the tx count include?

Got a reference for volume? Sol had either the most or second most active wallets.

I assure you that their status page is no indication of reliability. Ask the more serious node operators and they can tell you how frequently you actually encounter issues with the network.

Bitcoin can do that many transactions fairly easily with the changes we made for eCash. All the quadratic performance bottlenecks were removed.

Reply on news.ycombinator.com