Solana has suffered its second outage in a month, sending the price of SOL down by nearly 13% in the past 24 hours.
Solana co-founder Anatoly Yakovenko has tweeted an explanation:
“Durable nonce instruction caused part of the network to consider the block is invalid, no consensus could be formed.”
According to the Solana’s official documentation, “durable transaction nonce” refers to a mechanism on Solana addressing the typical short lifetime of a transaction block hash.
A bug in the feature caused nodes to generate different outputs, which resulted in a block hash mismatch and a failure for the network to reach consensus.
While Solana has been dubbed as an “Ethereum killer” due to it’s high scalability, the network has suffered outages on at least seven separate occasions since last year, including two denial of service attacks in September 2021. In April this year, the network was disrupted for almost 8 hours after it was overwhelmed by NFT mining bots.
Scalability > Security
Solana’s perceived “centralisation” has been a topic for debate since its inception. The network currently has only 1,800 validators, and has a Nakamoto coefficient of 19, which means that only 19 validators are needed to compromise the security of the blockchain.
However, it seems like the network’s higher degree of centralisation did not directly contribute to the recent disruptions, and also has its advantages.
“Solana definitely has fewer validators than other chains, but I would argue against how this makes it easier to deal with outages”, Solana FM’s Syamiir Othman told Blockhead.
“Take the outage in September 2021 for example. SOL had about 800 validators but consensus took about 17 hours. Today we have about 1800 validators and consensus took less than 5 hour”, he said.
The blockchain trilemma highlighted the trade-off between decentralisation, scalability, and security. In recent times, Ethereum’s inefficiencies have led to the development of more scalable but also more centralised networks.
For Solana FM head of operations Fathur Rahman, centralisation is necessary at the beginning to ensure scalability.
“When building anything, centralise first, decentralise later. You can’t scale fast at the start if you decentralise”, he said.