The Hidden Design Flaw That Contributed to Solana's 20-Hour Outage

Solana [$SOL$], a blockchain platform that has been gaining popularity recently, has been experiencing frequent outages. In fact, it has gone down about a dozen times already. So what is the reason for these outages? It turns out that it is all part of a massive design flaw, which we will try to break down in this article.

One of the key issues with Solana is that it, as well as a couple of other blockchain platforms, handles all its consensus on-chain. This means that all validator communications are done on the blockchain as a transaction, which inflates the transaction volume as well as the transactions per second [TPS]. In fact, less than 10% of the transactions on the network are actual transactions, such as performing a swap, sending or receiving, minting, etc.

The teal part of the bar graph in Solana's transaction volume is made up of validator messages or votes, which typically make up 90-95% of the total volume. Validators communicate and "agree" on what is proposed, such as updating balances after a swap. While this simplifies the process of reaching consensus, it also creates a number of problems.

One of the main issues is that when the Solana chain goes down, which happens all too often, validators need to go off-chain to begin discussing what to do. Where do they go? Discord! They then need 66% of validators to agree on a solution to get back up. However, what if many validators aren't available, such as when they are sleeping or just unavailable? This is what leads to a 20-hour outage of the network.

Relying on validators to be available and responsive requires extra work, and it all stems from the fact that Solana handles consensus on-chain. This is also an issue because it creates insane amounts of data for full nodes. While most nodes only store a portion of the blockchain history, since Solana uses an account-based model that only needs to store balances of accounts, every chain has many full history nodes, which store the entire history of the blockchain. In Solana's case, this means millions upon millions of unnecessary validator messages, which is part of why it requires a data center to run a full history node.

Hedera, another blockchain platform that operates the same way as Solana, also has crazy numbers for unnecessary transactions. In fact, the majority of transactions on Hedera are unnecessary, just like they are on Solana. $HBAR$, Hedera's native cryptocurrency, only does around 3-5 TPS on average.

While there are certainly other issues with Solana, such as faked TVL, misleading marketing, unaccounted for tokens, and being backed by the biggest scammer in crypto, understanding the design flaw of handling consensus on-chain is key to understanding why it experiences frequent outages. It is important to do your own research and not just follow the hype, otherwise you might end up investing in a scam or buying at the top. Once you understand how things work, you can make better decisions.