You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Explore Implementing a new API to allow users to replicate an operation batch rather than one operation. This might be beneficial for performance reasons if a customer batches operations before committing them via Raft.
The text was updated successfully, but these errors were encountered:
@metanet We are talking here about atomic batches, right? So one operation in the batch is committed iff all the other operations are committed.
If that is the case, then I would go with introducing a new type of operation, a BatchOperation, that stores the list of operations of a batch. The commitment status and log position of such a BatchOperation is then equal to the commitment status and position for all its containing operations. For applying a BatchOperation, we then would require the state machine to do this also atomically. As far as I see it, with that we would get atomicity for such a batch. What do you think?
Explore Implementing a new API to allow users to replicate an operation batch rather than one operation. This might be beneficial for performance reasons if a customer batches operations before committing them via Raft.
The text was updated successfully, but these errors were encountered: