How does SpatiX serve 2.5 billion users simultaneously? It relies on SpatiX's independently developed high concurrency capability built on the Spatial Intelligence platform – achieving "unlimited" service access capability through "limited" ground-based augmentation stations.
"High concurrency" is an internet terminology referring to the rapid access of a large number of users. For internet platforms, a lack of good high concurrency processing capability when faced with a large number of users accessing quickly can, at best, cause service interruptions and, at worst, lead to server crashes.
Moreover, interruptions in spatiotemporal service data can pose safety risks for "robots" such as operating drones and autonomous vehicles.
Therefore, SpatiX Spatial Intelligence has built high concurrency capabilities from three aspects: infrastructure, middleware layer, and distribution link, based on its own data processing characteristics.
Computing power is the foundation for internet platforms to achieve seamless expansion of business systems. SpatiX Spatial Intelligence adopts a distributed cloud computing architecture, which can stably expand user access capabilities at any time according to actual conditions.
Here, we first need to mention Alibaba Cloud's Elastic Compute Service (ECS) and Server Load Balancer (SLB).
Cloud Server ECS is an elastically scalable computing service that supports on-demand configuration of CPU and memory resources, as well as configuration upgrades.
Load Balancer SLB distributes traffic across multiple cloud servers to expand the external service capabilities of application systems and improve the availability of application systems by eliminating single points of failure.
Based on Cloud Server ECS and Load Balancer SLB, SpatiX Spatial Intelligence can, on the one hand, automatically increase the number of servers according to user needs to achieve rapid expansion of computing resources; on the other hand, it can distribute ultra-large traffic access requests in a short time to different servers for processing, performing "peak shaving" for instantaneous access from massive users.
This is like when the river water level rises, the dam automatically opens more gates, and the water flow can evenly flow to each gate.
SpatiX Spatial Intelligence features wide coverage, large user scale, and high real-time requirements, needing a high degree of flexibility to maximize efficiency.
In response to this situation, SpatiX has independently developed a high-performance distributed application framework, whose function is similar to that of a project manager.
After the system integrates the distributed architecture, it can divide a huge computing task into several small tasks and assign them to each computer. The computer cluster improves efficiency through parallel processing and completes large-scale information processing tasks under unified management and control.
At the same time, the distributed architecture adopts a self-designed transmission protocol, enabling high-speed real-time transmission of data streams between nodes. An ordinary cloud server of SpatiX can support the transmission of 10,000 data points.
Through functions such as task scheduling and management, computer health status checks and second-level switching, load balancing, route selection, and gray-scale upgrading, the distributed architecture quickly realizes a distributed processing system cluster architecture with high throughput, low latency, and high availability in application systems.
Based on SpatiX's distribution business characteristics and referring to the high availability design ideas of internet network structures, SpatiX's distribution data link model is divided into four levels: core processing, amplification, aggregation, and access, with the ability to "turn a complex big problem into many simple small problems".
The core processing layer mainly processes raw satellite observation data received in real-time from the ground reference station network, calculates error correction information, and generates meter-level, centimeter-level, and millimeter-level differential data products.
The amplification layer is the "agent" of the core processing layer. After the core processing layer generates differential data products, it hands them over to the amplification layer for "agency" to reduce the load on the core processing layer and support multi-level amplification.
The aggregation layer is the "intermediary" between the access layer and the core processing layer. It aggregates differential product data before user terminals receive it, thereby maximizing the reuse of processing links, improving the efficiency of computer resource utilization, and reducing service pressure.
The access layer provides functions such as service access and authentication for user terminals.
In addition to the core processing layer, the amplification layer, aggregation layer, and access layer can all achieve horizontal expansion, laying the foundation for massive users to obtain real-time, stable, and efficient SpatiX Spatial Intelligence services.
SpatiX's distribution link has undergone disaster recovery and fault drills, and through continuous accumulation of testing experience in various extreme scenarios and repeated automated testing, the high availability of the entire link is guaranteed.