cicv data center & proprietary cloud platform is the first high-level data center built around the intelligent network connection industry in china. the data center is built according to the national standard a grade and leads the industry in building structure, electrical system, air conditioning system, water replenishment and fuel storage. the data center is a data center relying on the intelligent network connection industry cloud as the core, and several cloud platforms have been built internally to support the business covering the whole industry chain of intelligent network connection, mainly including intelligent network connection r&d cloud, cloud control infrastructure platform edge cloud, cloud control infrastructure platform regional cloud, cloud control infrastructure platform central cloud, intelligent network information security cloud, cicv-data high precision map cloud, and aicc operating system cloud.
the infrastructure of each sub-cloud platform in the data center follows the construction guidelines of layered decoupling and cross-domain scaling, while five construction standards are set in the technical selection of each cloud platform.
(1) stability: the core is to have financial-level disaster recovery capability, to support disaster recovery, multi-live and disaster recovery in two locations and three centers, and to have the systematic support capability to realize disaster recovery and backup in iaas and paas. the architecture should support the deployment of super large-scale nodes of tens of thousands of levels, and the capability should pass the national credible cloud certification.translated with www.deepl.com/translator (free version)
(2) decoupling: mainly the realization of layered decoupling capability. first of all, to achieve hardware and software decoupling, the cloud platform physical equipment needs to support the deployment of multiple chips and multi-brand x86 servers, iaas layer software to achieve a cloud platform to support the mixing of equipment based on different products, while the iaas layer software to provide standardized interfaces open to developers; secondly, to achieve iaas and paas, saas layer decoupling, support commercial paas suite, open source paas software products are deployed to iaas platform, and saas is supported to be deployed directly to iaas platform; through the deep decoupling of software and hardware as well as iaas, paas and saas, the locking of vendors is effectively avoided.
(3) scalability: mainly the realization of cross-domain scalability. first of all, it needs to support the scalability of a single cloud platform within 50km across physical facilities, and the cloud platform needs to have the architectural capability of a cloud with multiple regions to support the rapid expansion of computing clusters. secondly, the big data software inside the cloud platform should support the scheduling of arithmetic clusters across regions and support the logical and physical isolation of arithmetic clusters. finally, support the self-built cloud platform and the public cloud to expand interoperability, to provide capacity support for machine learning training scenarios to achieve rapid training models in a short period of time.
(4) security: it has the ability to support each business platform to pass the level 3 certification of equal security and support national level security assessment. for the high reliability requirements of the cloud control infrastructure platform, the iaas layer of the platform needs to have the ability to support the completion of the assessment of level 4 of the equal protection.
(5) intelligent: cloud management system supports the unified management of a cloud multi-region, multi-cloud management, cloud edge integrated management, in operation and maintenance can effectively realize the unified management of mainstream cloud platform, centralized intelligent operation and maintenance.