back
1 comments
>The most popular mechanism, which guarantees consistency at minimal cost to availability, is to use the Paxos or Raft consensus protocols to enforce consistency across multiple replicas of the data.

I cannot believe that someone with his bio wrote that sentence. He seems to be using the word "consistency" to mean durability, which is fine (although incomplete) in the context of CAP theorem, but "durability" is completely different from the meaning of consistency in "eventual consistency" where it actually refers to linearizability.

Also in the comments he seems to be confusing linearizability with transaction total order. I don't believe linearizability implies a total order of transactions, unless you also have serializability+linearizability like Spanner (which essentially makes every transaction a single operation, on which linearizability can then apply).