Jepsen @jepsen@jepsen.io
- GitHub
- https://github.com/jepsen-io
Breaking distributed systems, one fault at a time.
Again, this is just design review--I can't speak to anything about Replicache's implementation--but these properties are theoretically achievable! There's precedent in both Bayou and Eventually-Serializable Data Services, both from 1996! I'm kinda curious what kind of cross-pollination was going on there.
https://groups.csail.mit.edu/tds/papers/Lynch/podc96-esds.pdf
Last fall, Replicache commissioned Jepsen to take a look at their design docs, and we wrote a bit about how it works: https://replicache.dev/jepsen.html.
They're doing something kinda like Bayou for mobile apps: local, speculative JS transactions, which are available even offline, replicated asynchronously. Those transactions may be reordered/re-executed with varying effects, but always preserve causal ordering; eventually they stabilize on a fixed serial order.
Hi Fediverse! You can follow this instance for Jepsen-related news--new analyses, project releases, etc.
- GitHub
- https://github.com/jepsen-io
Breaking distributed systems, one fault at a time.
