Syncing data across user devices for distributed collaboration

Syncing data across user devices for distributed collaboration

Slides from a presentation given on 12 July 2019 at the Hydra distributed computing conference, St. Petersburg, Russia.
https://martin.kleppmann.com/2019/07/11/hydra-distributed.html

Abstract:

Google Docs and similar web apps are very convenient if several people need to work together on a document: all users can simultaneously edit the shared document without having to send files back and forth by email. From a distributed systems point of view, real-time collaboration is a replication problem: each user has a replica of the shared document on their device, which they can modify locally, and any changes are sent over the network to their collaborators.

Users may edit a document while offline, causing their replica to become temporarily inconsistent with the others. As they re-synchronise with others, conflicts need to be resolved so that all collaborators end up in a consistent state. Conflict-free Replicated Data Types (CRDTs), first proposed in 2011, are increasingly being used to merge different users’ versions of a document in collaborative software.

This talk will give an update on the latest research on CRDTs, and our open source implementation Automerge. We will also discuss efforts to build “local-first software”, a new generation of collaborative software, on top of these technologies.

0d4ef9af8e4f0cf5c162b48ba24faea6?s=128

Martin Kleppmann

July 12, 2019
Tweet