Sync Engine Comparison for Local-First Web Applications : An Empirical Study
Shiferaw, Ermyas Fekadu (2026)
2026
Master's Programme in Computing Sciences and Electrical Engineering
Informaatioteknologian ja viestinnän tiedekunta - Faculty of Information Technology and Communication Sciences
Hyväksymispäivämäärä
2026-06-02
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tuni-202606016679
https://urn.fi/URN:NBN:fi:tuni-202606016679
Tiivistelmä
Web development is increasingly adopting the local-first approach, in which data is stored and processed on the user’s device to enable responsive interactions and reliable offline use. In this architecture, the sync engine is a core component responsible for maintaining consistency between local and remote replicas across devices and servers. Although the theoretical foundations of data replication and conflict resolution are well established, direct comparative research on different families of sync engines, as well as on developers’ experiences of using them, remains limited.
This thesis addresses that gap by empirically comparing six modern sync engines: Zero, PowerSync, ElectricSQL, Yjs, Automerge, and ShareDB. The study uses a two-phase mixed-methods design. First, a technical audit evaluates the engines against the Seven Ideals of Local-First Software to identify their architectural trade-offs. Second, an exploratory survey of 13 developers captured real-world perspectives on usability, integration effort, ecosystem fit, and operational complexity.
The findings show that no single engine fully satisfies all local-first ideals out of the box. Relational engines show strong support for structured data management and connecting to standard backends, but they struggle with complex simultaneous edits. On the other hand, document-based Conflict Free Replica Data Type (CRDT) engines are well suited for real-time collaboration but come with a steep learning curve, higher memory usage, and difficult debugging. Ultimately, this thesis concludes that choosing a sync engine is a practical engineering decision to match use case, requiring system architects to carefully match the engine family to the specific functional needs of their application.
This thesis addresses that gap by empirically comparing six modern sync engines: Zero, PowerSync, ElectricSQL, Yjs, Automerge, and ShareDB. The study uses a two-phase mixed-methods design. First, a technical audit evaluates the engines against the Seven Ideals of Local-First Software to identify their architectural trade-offs. Second, an exploratory survey of 13 developers captured real-world perspectives on usability, integration effort, ecosystem fit, and operational complexity.
The findings show that no single engine fully satisfies all local-first ideals out of the box. Relational engines show strong support for structured data management and connecting to standard backends, but they struggle with complex simultaneous edits. On the other hand, document-based Conflict Free Replica Data Type (CRDT) engines are well suited for real-time collaboration but come with a steep learning curve, higher memory usage, and difficult debugging. Ultimately, this thesis concludes that choosing a sync engine is a practical engineering decision to match use case, requiring system architects to carefully match the engine family to the specific functional needs of their application.