Get the latest tech news
CRDTs #2: Turtles All the Way Down
This is the 2nd post in a series of 4 posts I'm doing on CRDTs. Please see the intro post for context. Modern distributed systems often seem to rest on an stack…
They promise convergence of state across machines without requiring perfect clocks, global operation ordering, or causal message delivery ... and they do it with math. The basic idea is this: even after all updates have been issued, nodes can pass an item back and forth as a "hot potato" indefinitely, and never converge despite communicating infinitely often! Typically, op-based CRDT designs assume that the log at each site is "played" (eagerly or lazily), by executing the ops in their causal partial order to materialize the local state.
Or read this on Hacker News
