[SOIN Technology News] Recently, SOIN Technology and Shenzhen Yashan Technology Co., Ltd. (hereinafter "Yashan Technology") formally entered into a comprehensive, in-depth cooperation on the YashanDB database product. The two parties will deepen collaboration around the domestic enterprise-grade database across Xinchuang (IT application innovation) AI servers, joint industry solutions and ecosystem co-building, further strengthening SOIN's "Enterprise AI Digital Foundation" strategy anchored by its SOINAWE series of enterprise-grade AI inference servers.
Domestic Database on the Server: Completing the Autonomous, Controllable Foundation
Earlier, SOIN Technology officially launched the SOINAWE-RVE1300 enterprise-grade high-performance AI inference server, building its Enterprise AI Digital Foundation on the self-developed AWE engine plus a Xinchuang ecosystem. One of the core outcomes of this cooperation is that this AI inference server, built for enterprise-critical scenarios, will deeply integrate YashanDB — giving customers, alongside high-performance AI inference, a fully autonomous, controllable domestic database foundation with a high degree of Oracle compatibility and full Xinchuang compliance, delivering "compute + data + AI" as an integrated package.
100% Autonomous Kernel, Rooted in Source Innovation Led by an Academician
YashanDB is developed by the Shenzhen Institute of Computing Sciences (SICS, "Deep Computing Institute"), one of Shenzhen's ten foundational research institutions and the product of 100% state-owned investment. Its creator, Academician Wenfei Fan, is a rare "grand slam" scholar spanning both database theory and systems; the team brings together core R&D talent from Huawei, IBM and Alibaba, with R&D accounting for over 70% of a 500-plus-person team. YashanDB was built entirely from the first line of code, with no secondary development on any open-source database, and its kernel self-development rate reaches 100% (verified by an authoritative body). It also embeds Academician Fan's four original theories — bounded computing, cross-model computing, approximate computing and parallel scalability — turning academic theory into usable systems and addressing the "chokepoint" problem at its root.
Over 98% Oracle Compatibility, Near-Zero-Change Smooth Replacement
For enterprises advancing Xinchuang substitution, Oracle compatibility is a hard requirement. YashanDB achieves four-layer compatibility across architecture design, functional features, ecosystem interfaces and user habits, with Oracle compatibility exceeding 98%: it is compatible with 26+ data types, 50+ stored-procedure capability points, 230+ system package functions, 130+ built-in functions, 200+ dictionaries and 60+ dynamic views; it is compatible not only with syntax, semantics and PL/SQL but also delivers "performance compatibility." In real production migrations, one system achieved a measured Oracle compatibility of 99.76%, replacing with almost zero application changes.
Full Xinchuang Adaptation, Embedded in the Domestic Software-Hardware Ecosystem
YashanDB fully supports the Xinchuang technology ecosystem, having adapted to mainstream domestic CPUs, operating systems, middleware, servers and cloud platforms, and completed compatibility certification with 100+ channel and strategic partners and 400+ upstream/downstream industry players. For SOIN, this means the database foundation carried by the RVE1300 server can be seamlessly incorporated into a full-stack Xinchuang list from chip to application, providing compliance assurance for the migration of government and enterprise customers' critical systems.
Financial-Grade High Availability and Shared Cluster, on Par with Oracle RAC
For high-end core transaction scenarios, YashanDB launched the fully self-developed shared cluster YAC (YashanDB for Cluster), achieving concurrent multi-node read/write with strong consistency and providing an Oracle RAC-equivalent replacement with "unchanged architecture, unchanged application code, unchanged O&M habits." Its financial-grade high availability reaches RPO=0 and RTO<10 seconds, with 4-node TPC-C performance of 6.18 million tpmC and a linear scaling ratio above 0.7, while supporting two-site three-center and multi-site multi-center active-active deployments to meet the stringent 7x24-hour requirements of bank and securities core systems.
Secure, Reliable and Compliant, Certified by National-Level Assessments
Security is the bottom line of a database. YashanDB passed the authoritative "Security and Reliability Assessment" (Level II) by the China Information Technology Security Evaluation Center and holds EAL4+ network-security certification, Level 4 classified protection, commercial cryptography and classified-security qualifications. It supports the SM2/SM3/SM4 national cryptographic algorithms, tablespace-level and column-level transparent encryption (TDE), a zero-trust security architecture, separation of powers (three-authority) and label-based access control (LBAC), building end-to-end protection from storage and transmission to access.
AI-Ready Data Foundation for the Agent Era
For the AI era, YashanDB provides a multi-model converged data foundation that unifies relational, vector, GIS, graph and JSON data and supports precision-guaranteed relational-vector hybrid queries. It also pioneered the kernel-native "data sandbox" (millisecond-level branch cloning — managing data like Git) and the enterprise-grade agent governance solution YashanClaw, resolving the trust and isolation challenges of "how enterprises can safely use AI." This aligns closely with the positioning of SOIN's SOINAWE series of AI inference servers — database and compute together become the reliable foundation for enterprise AI adoption.
Exceptional Cost-Effectiveness and One-Stop Smooth Migration
Beyond technical capability, YashanDB also excels in total cost of ownership: low License/MA fees, low application-retrofit cost and low maintenance and labor investment, with reusable hardware and fewer server nodes. Combined with a one-stop database migration platform offering automated assessment, application adaptation and rewriting, dual-track parallel running, gray-cutover and rollback verification, it drastically lowers the Xinchuang substitution threshold. In financial benchmark cases, a top-tier brokerage cut overall TCO by 60%, and a securities firm migrated 16.7 TB of data, 269 stored procedures and 224,000 lines of code with near-zero changes; its capabilities already serve the Digital Currency Research Institute of the PBOC, Kazakhstan's Kaspi Bank (TCO savings over 30%), provincial rural commercial banks, and the power, transportation and airport sectors.
Building Together, Lowering the Barrier to Xinchuang and AI Adoption
Through this comprehensive, in-depth cooperation, SOIN and Yashan combine the high-performance AI inference capability of the RVE1300 server with YashanDB's domestic database foundation, giving government and enterprise customers an integrated digital foundation that is "out-of-the-box, autonomous and controllable, smoothly substitutable and AI-oriented." Drawing on their respective strengths in compute hardware and foundational software, the two parties will continue to launch joint solutions across finance, government, energy and transportation, helping more critical business systems move safely and smoothly toward Xinchuang and intelligence.