In large-scale Industrial Internet of Things (IIoT) environments (e.g., intelligent manufacturing, energy management), cross-domain sharing of high-value data encounters critical challenges, including third-party attacks, latency bottlenecks, and inconsistencies arising from untrusted sources. Existing solutions suffer from efficiency limitations due to the decoupling of authentication and consensus mechanisms. To address these issues, we propose IDSF, a cross-domain trustworthy data sharing framework for large-scale IIoT, integrating Decentralized Identifiers (DID) with an enhanced Practical Byzantine Fault Tolerance (PBFT) algorithm. Specifically, a lightweight DID-based authentication protocol is designed to eliminate third-party dependencies via a local hash table mechanism, significantly reducing latency and improving energy efficiency. Furthermore, we construct a dynamic reputation evaluation model for IIoT devices and design a reputation-threshold-controlled PBFT (RPBFT) algorithm to effectively balance Byzantine fault tolerance with low-latency requirements. Extensive simulations demonstrate that the proposed authentication protocol surpasses existing methods in both efficiency and energy consumption. Compared to traditional PBFT, the RPBFT algorithm reduces consensus delay by 34.8% and improves average throughput by 25.2%. These results validate the framework’s capability to support scalable IIoT data sharing, enabling thousand-node deployment in smart factory environments.