A Scalable and Resilient Protocol for Synchronous Collaboration

Zhengqi Wu,Hui Li,He Bai,Xinyuan Pei

DOI: https://doi.org/10.1109/cscwd61410.2024.10580450

2024-01-01

Abstract:Synchronous collaboration has emerged as an indispensable service in maintaining data consistency across nodes within extensive distributed systems. Most distributed systems rely on synchronization protocols to access and coordinate with the data of participants. However, existing synchronization protocols are limited to full synchronization patterns, neglecting the need for partial data synchronization in pub-sub scenarios. Furthermore, they exhibit deficiencies in the design of their notification mechanisms, particularly when it comes to concurrent updates and packet loss handling. In this Paper, we propose MINSync to address the synchronization needs of diverse scenarios over Multi-Identifier Networking (MIN), a novel network architecture with enhanced security and content retrieval efficiency. MINSync offers support for partial and full data synchronization. Furthermore, it facilitates both pull-based and push-based notification mechanism. The protocol also deploys a new forwarding strategy to efficiently handle concurrent updates. The experimental results demonstrate that MINSync consistently reveals excellent scalability as the number of consumers, subscriptions, and streams increases. It also has lower synchronization delay in concurrent update and packet loss situations compared to PSync.