High Responsiveness for Group Editing CRDTs

Abstract

Group editing is a crucial feature for many end-user applications. It requires high responsiveness, which can be provided only by optimistic replication algorithms, which come in two classes: classical Operational Transformation (OT), or more recent Conflict-Free Replicated Data Types (CRDTs). Typically, CRDTs perform better on downstream operations, i.e., when merging concurrent operations than OT, because the former have logarithmic complexity and the latter quadratic. However, CRDTs are often less responsive, because their upstream complexity is linear. To improve this, this paper proposes to interpose an auxiliary data structure, called the identifier data structure in front of the base CRDT. The identifier structure ensures logarithmic complexity and does not require replication or synchronization. Combined with a block-wise storage approach, this approach improves upstream execution time by several orders of magnitude, with negligeable impact on memory occupation, network bandwidth, and downstream execution performance.

Real Differences between OT and CRDT under a General Transformation Framework for Consistency Maintenance in Co-Editors

“…For example, the TTF solution was reported to be outperformed by Logoot and RGA for a factor up to 1000 in [4]. This 1000-times-gain claim was highly remarkable and widely cited as an experimental evidence for CRDT's performance superiority over OT [4,5,8,45,81] (also reflected in footnote 7).…”

“…The proposed solution was to add one extra binary tree to speed up the local search process. With this extra tree, together with its additional complication and costs, the time complexity of this search process could be reduced to O(log(Ct)), which improves the upstream execution time by "several orders of magnitude" as claimed in [8] 7 . For comparison, OT solutions commonly have the complexity O(1) for the whole process of handling a local operation.…”

“…A more recent paper [8] tried to address the CRDT performance problem in the upstream processing (i.e. processing local operations).…”

“…While WOOT variations and RGA are different from each other in concrete representations of internal object sequences and operations, they have one thing in common: the use of tombstones to 7 In [8], the authors acknowledged "upstream execution ̶ and thus responsiveness of CRDT algorithms often performs poorly," but still claimed "downstream execution of CRDT algorithms is more efficient by a factor between 25 to 1000 compared to representative OT algorithms", quoting the evaluation results in [4], where the referred OT algorithm is the TTF solution [42], which is examined in Section 5.2. Object sequence @s, o0, o1, …, on, @e where ox is an object corresponding to a character in external document or a tombstone; @s and @e are special objects marking the start and end of the object sequence.…”

“…Recent years have seen OT being widely adopted in industry products as the core technique for consistency maintenance, ranging from battle-tested online collaborative richtext editors like Google Docs 1 [11], to emerging start-up products, such as Codox Apps 2 .A variety of alternative techniques for consistency maintenance in co-editors had also been explored in the past decades [14,16,18,41,42,72]. One notable class of techniques is CRDT 3 (Commutative Replicated Data Type) for co-editors [4,5,8,25,32,37,39,40,41,45,47,48,79,80,81]. The first CRDT solution for plain-text co-editing appeared around 2006 [40,41], under the name of WOOT (WithOut Operational Transformation).…”