Scientific data are posing new challenges to data management due to the large volume, complexity and heterogeneity of the data. Meanwhile, scientific collaboration becomes increasingly important, which relies on integrating and sharing data from distributed institutions. In this demo, we present SciPort, a Web-based platform on supporting scientific data management and integration based on peer-to-peer architectures, where researchers can easily collect, publish, and share their complex scientific data across multi-institutions. SciPort provides a general metadata based data model to capture the context description of experiments and link experiment data into comprehensive metadata documents, and supports a hierarchical organization of the overall data space for data browsing. SciPort takes two alternative "peer"-to-"peer" (or peer-database-to-peer-database) based approaches to integrate scientific data: pure peer-to-peer architecture and central server based peer-to-peer architecture. The later provides a virtual view of all published data from multiple local sites and supports complex queries with XQuery. The system provides a unified framework for adaptable architectures and customizable schemas, and supplies comprehensive tool set to manage and share scientific data. SciPort was first prototyped in Siemens Corporate Research, and now becomes a mature product and has been successfully used in both biomedical research and clinical trials for scientific research communities.