Read on Twitter
Klaytn, the South Korean blockchain project, had an outage that lasted 13 hours.
There are two bits of crucial information here that most mainstream coverage missed and will continue to miss.
Klaytn Blockchain Prevented from Generating Blocks — https://www.btcwires.com/c-buzz/klaytn-blockchain-prevented-from-generating-blocks/
There are two bits of crucial information here that most mainstream coverage missed and will continue to miss.
Klaytn Blockchain Prevented from Generating Blocks — https://www.btcwires.com/c-buzz/klaytn-blockchain-prevented-from-generating-blocks/
Before this, EOS had a similar outage.
So did Stellar.
The common thread between all these platforms is that they are all based on classical consensus protocols.
So did Stellar.
The common thread between all these platforms is that they are all based on classical consensus protocols.
Classical consensus protocols were very well understood when Satoshi was looking to devise an internet scale cryptocurrency. He rejected them because they are all fragile.
I've heard @VladZamfir mention many times that he fully expects ETH2 to experience liveness problems. This is because ETH2 is based on a classical algorithm, not very different from EOS.
Classical protocols are fragile, and they also accommodate only a tiny number of participants. (As an aside, that's also why Libra, based on the classical HotStuff protocol co-authored by my student @Tederminant, aspires to have at most 100 validators).
So, my Lesson #1 from the Klaytn outage.
These liveness issues are to be expected with the crop of PoS coins out there that are based on classical protocols.
These liveness issues are to be expected with the crop of PoS coins out there that are based on classical protocols.
The avalanche protocol is different. It's far more robust. Does not require precise agreement on membership. Nodes can disagree on which nodes they believe to be up or down, and still make sound decisions.
My Lesson #2 from the Klaytn outage.
It doesn't matter who is behind a chain and what kind of experience they have with running centralized services.
It doesn't matter who is behind a chain and what kind of experience they have with running centralized services.
Blockchains are fundamentally different. You could be Kakao, or Telegram, and still watch your blockchain falter. Your expertise with datacenters and admins on call is absolutely useless.
Blockchains are very resilient or prone to failure. The difference depends not at all on operational skills. In fact, the very idea of "operating" a decentralized chain should be a red flag. What matters is the ability to design self-tending, robust protocols.
