I get it now. It is all thanks to separating the operation as
1. agreeing on WAL,
2. executing after waiting all WAL agreements stabilizing.
There is a separation between the two. This can introduce waits (which could be theoretically unbounded for EPaxos but Tempo and Accord bounds this).
Maybe in some sense this is in the same spirit of Calvin's serializing WAL first and executing it later. But Accord avoids a single entity doing the serialization and achieves it via a leaderless Paxos approach which uses dependency graphs for WAL serialization, and later using another protocol for WAL execution after settlement.
> Maybe in some sense this is in the same spirit of Calvin's serializing WAL first and executing it later
Yes, it is very similar in effect, with the absence of a global log conferring some important differences with pluses and minuses. I expect to offer interactive transactions eventually, for instance, which may or may not be difficult for Calvin, and faults on a “log shard” are isolated with Accord, but have global impact with Calvin. But snapshot queries are much easier with Calvin.
I get it now. It is all thanks to separating the operation as 1. agreeing on WAL, 2. executing after waiting all WAL agreements stabilizing.
There is a separation between the two. This can introduce waits (which could be theoretically unbounded for EPaxos but Tempo and Accord bounds this).
Maybe in some sense this is in the same spirit of Calvin's serializing WAL first and executing it later. But Accord avoids a single entity doing the serialization and achieves it via a leaderless Paxos approach which uses dependency graphs for WAL serialization, and later using another protocol for WAL execution after settlement.