I have been reading "Distributed Transactional Systems Cannot Be Fast" by Didona et al (arxiv.org/pdf/1903.09106.pdf) recently and find it fascinating! Assuming the proof holds, it's basically saying that applications either get fast reads or transactions, but never both.

However, the paper does not talk about CRDTs which seem to be able to solve some part of this problem by providing fast reads and strong eventual consistent writes. What other work in this area that I should be reading?

Follow

@penberg “distributed” in this sense meaning “not a shared memory system”?

Sign in to participate in the conversation
Mastodon

General topic personal server.