Multi-core accelerated operational transformation for collaborative editing

Abstract

This article proposes a parallel operational transformation (OT) algorithm for collaborative editing. OT maintains the eventual consistency of replicated data in optimistic way, allowing users to manipulate the shared document simultaneously. It has been the first choice for most collaborative applications. However, existing approaches must keep the number of operations generated in a session small so that it can provide a decent responsive time. The multi-core/many-core architectures are becoming pervasive in recent years. Unfortunately, there is no prior work which has explored accelerating operational transformation algorithms with available computation power. We present a lock-free operation history which are accessed by a batch of remote operations at the same time. Moreover, a data parallel computation model is constructed to accelerate the integration of local operations. To the best of our knowledge, this is the first parallel OT algorithm. Experimental results show our proposed algorithm outperforms the stat-of-art algorithms for collaborative editing.