Solution Architecture

The platform is divided from bottom to top into four layers: hardware-equipment layer, infrastructure-resource layer, base-software-resource layer, and resource-management interface:

Hardware-equipment layer:Supporting hardware of the infrastructure resource pool mainly includes x86 physical servers, distributed NAS devices, and network devices. Hardware of the virtualization pool and distributed physical pool is all x86 physical servers; x86 compute servers and x86 compute-storage servers of different performance are configured by server functional use (management nodes, compute nodes, compute-storage nodes), and compute clusters are built in a software-defined way (virtualization software, container engines, distributed block-storage software, distributed file systems).

Infrastructure-resource layer:Through virtualization and distributed technologies, pool hardware devices, build compute and storage clusters in different resource forms such as VMs, containers, and physical nodes, realize logical abstraction, encapsulation, scheduling, and metering of resources, and then build resource pools.

Base software resources:Deploy base software on infrastructure resources, turn resources into services and capabilities, and provide software-resource services such as distributed computing environments, data-storage support environments, and middleware. The distributed computing environment mainly provides stream, batch, and in-memory computing framework engines, plus data-mining and machine-learning environments. The data-storage support environment, mainly according to the design of the data-storage and service subsystem, provides distributed table systems, distributed relational databases, distributed analytical databases, NoSQL databases, primary distributed file/object storage, secondary distributed file storage, and backup/archive services. Middleware includes messaging, cache, web, visualization, GIS, and other software resources and services.

Resource-management interface:Different types of infrastructure resources (VMs, containers, distributed physical nodes (compute, storage), distributed NAS/object-storage devices, etc.) need an overall management view; the same type of infrastructure resources has different technical implementations/architectures, and heterogeneous same-category resources need a unified resource-lifecycle management process. Therefore, a resource-management interface needs to be deployed to unify resource management, scheduling, and resource monitoring; the business-monitoring system uniformly provides resource services through this interface.