Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Extending JSON CRDTs with Move Operations

Liangrun Da, Martin Kleppmann

TL;DR

An algorithm for move operations in a JSON CRDT that handles the interaction with concurrent non-move operations, and uses novel optimisations to improve performance is introduced and integrated into the Automerge CRDT library.

Abstract

Conflict-Free Replicated Data Types (CRDTs) for JSON allow users to concurrently update a JSON document and automatically merge the updates into a consistent state. Moving a subtree in a map or reordering elements in a list within a JSON CRDT is challenging: naive merge algorithms may introduce unexpected results such as duplicates or cycles. In this paper, we introduce an algorithm for move operations in a JSON CRDT that handles the interaction with concurrent non-move operations, and uses novel optimisations to improve performance. We plan to integrate this algorithm into the Automerge CRDT library.

Extending JSON CRDTs with Move Operations

TL;DR

An algorithm for move operations in a JSON CRDT that handles the interaction with concurrent non-move operations, and uses novel optimisations to improve performance is introduced and integrated into the Automerge CRDT library.

Abstract

Conflict-Free Replicated Data Types (CRDTs) for JSON allow users to concurrently update a JSON document and automatically merge the updates into a consistent state. Moving a subtree in a map or reordering elements in a list within a JSON CRDT is challenging: naive merge algorithms may introduce unexpected results such as duplicates or cycles. In this paper, we introduce an algorithm for move operations in a JSON CRDT that handles the interaction with concurrent non-move operations, and uses novel optimisations to improve performance. We plan to integrate this algorithm into the Automerge CRDT library.
Paper Structure (16 sections, 4 figures)

This paper contains 16 sections, 4 figures.

Table of Contents

  1. Introduction
  2. The Core Mechanics of Automerge
  3. The Moving Algorithm
  4. Validity of Move Operations
  5. Performance Optimisation
  6. Further Optimisations
  7. Batch Updating
  8. Lifecycle Tracking
  9. Evaluation
  10. Convergence Complexity and Performance
  11. Overhead Caused by Move Support
  12. Correctness Testing
  13. Related Work
  14. CRDTs for Trees
  15. CRDTs for Lists
  16. ...and 1 more sections

Figures (4)

  • Figure 1: Initially, nodes $A$ and $B$ are siblings. Replica 1 moves $B$ to be a child of $A$, while concurrently replica 2 moves $A$ to be a child of $B$. Boxes (a) to (d) show possible outcomes after the replicas have communicated and merged their states: (a) A and B form a cycle; (b) concurrently moved subtrees are duplicated; (c) Replica 2's move is ignored; (d) Replica 1's move is ignored. Figure from kleppmann2021highly.
  • Figure 2: An example JSON document with its internal OpSet
  • Figure 3: Convergence time of two actors that diverge by move operations
  • Figure 4: Convergence time of two actors that diverge by non-move operations