Solana publishes review on its network outage, thanks community
On September 14, the Solana network went down due to a sudden increase in the network load to over 400,000 transactions per second and this was fixed after around 17 hours of downtime and due to the efforts of all the validators within the network.
This had caused a big fall in the price of SOL tokens and it was believed that it could be the tip of a very big iceberg which the Solana community was living with. Such critics tend to conveniently forget that even much larger networks like Ethereum had many high downtimes, lasting days, at the time when the networks had just begun to start their journey, right where Solana is right now. The Solana community has published a review and explanation for the downtime now and has promised to do an even further root cause analysis in due course of time.
It has now been revealed that the increase in traffic was due to a denial of service attack launched on the network when the Grape Protocol launched its IDO on the Raydium network. This caused flooding of the network and the validators could not confirm the current status of the blockchain which led to new blocks getting accumulated without confirmation. It says that efforts were made to restart and recover the network but due to the piling up of the blocks, the validators began to lose memory very quickly and could catch up with the rest of the network in time. So it was decided to do hard work and engineers around the world worked on the fix and then there was an update and restart of the network from across more than 1000 validators.
This is one of the biggest advantages of decentralization and blockchain as there is not a single point of failure and the network does not need to look up to that single source to come back up. A classic example of centralization would be AWS as if the servers there crash, then all the applications and the networks would need AWS to come back up and the others would not have a choice but to wait for that to happen. But with a decentralized network, the validators do not need to wait for a signal from a central source and they can get together, coordinate, and fix the network if needed and it enough if around 80% of the network to be up at any point of time for the blocks to be confirmed and the network to work as normal. The Solana community was able to show such decentralization in action though it would be well served to remember the lessons learnt and ensure that such outages do not happen again as it could lead to loss of trust from users and developers as well.