mcms
Joined 12 years ago
Joined 12 years ago
And I don't think OLTP/OLAP is about LWTs. I think they are just talking about Scylla's ability to process normal updates in a timely manner while doing OLAP queries.
And you're right about ACID transactions. Multiple rows across two or more partitions can not be updated atomically, even with LWTs, but it is possible for a single partition.
It mentions netmap, DPDK, OpenOnload, etc.
I find over-relying on encapsulation more harmful than useful nowadays specially if you are going to write scalable software that are inherently distributed. For example, hiding accessing a database behind a simple getter function makes another programmer ignore performance implication and other issues that may arise.
From what I understand, transaction A can still be replicating intents and transaction B, not knowing that coordinator for transaction A is still at work, start a recovery process. This makes transaction A abort which could be prevented by waiting for intents of A to be successfully replicated.