In the context of accelerating global smart city development, a large city in the Middle East successfully upgraded its public safety video surveillance system (CCTV) to an intelligent platform by deploying the ZStack Cloud platform. This platform, with elastic computing, GPU intelligent scheduling, and high-availability architecture at its core, supports the processing of thousands of high-definition video streams and AI-based analysis, helping the city’s security system evolve from “passive monitoring” to “proactive intelligent defense.”
This provides a replicable benchmark for the cloud transformation of global smart cities.
Challenge: Traditional Architecture Struggles with Smart Upgrade Demands
To advance its smart city strategy, a large city in the Middle East, led by the government, built a city-wide public safety video surveillance system (CCTV). The system integrates resources from multiple departments such as public security, transportation, and urban management, and is responsible for key functions like road surveillance, traffic enforcement, and security monitoring in critical areas. It serves as the digital backbone for urban security coordination and emergency response.
With the rapid increase in demand for HD video and AI analysis, the client needed to upgrade their system via cloud technology to overcome the limitations of traditional physical architecture, constructing an efficient, elastic, and intelligent security cloud platform.
The customer’s original system used a distributed physical server deployment, facing three key issues:
- Low resource utilization: Video stream processing and AI task loads were uneven, leading to high hardware costs with both performance redundancy and insufficiency.
- Limited scalability: The system was unable to quickly meet the demands for new AI tasks like facial recognition or license plate recognition, with performance limited by single-machine GPU resources.
- High operational complexity: Isolated systems across different departments and fragmented security policies made it difficult to meet the stringent requirements for high availability and long-term stability for a national-level platform.
ZStack Cloud Platform: Building a City-Level Intelligent Video Hub
In response to the client’s needs, ZStack designed a multi-data center cloud architecture, focusing on elastic computing, intelligent scheduling, and hybrid storage, to build a high-performance security cloud platform:
1. Efficient Resource Pooling, Unleashing Computing Power Potential
- High-performance server clusters were deployed across two locations, creating dual computing resource pools managed centrally by ZStack for unified CPU/GPU resource management.
- GPU passthrough technology was used for AI task-specific GPU allocation, improving the response speed of models like license plate recognition and behavioral analysis by 40%.
- A hybrid architecture combining centralized storage and local disk helped meet the demands for high-throughput video streams and cost-efficient archiving, improving IO performance by 50%.
2.High Availability Architecture Supporting Critical Business Operations
- Dual management nodes and computing nodes were deployed together, ensuring 24/7 uninterrupted platform operation with the system running stably for over 500 days without any downtime.
- The platform supports hot upgrades and security hardening, allowing patching without requiring system downtime, meeting government system compliance requirements.
3. Agile Delivery and Centralized Management
- The lightweight cloud platform was deployed within one week, shortening the delivery cycle by 50% compared to traditional solutions.
- Multi-tenant isolation and unified permission management enabled decentralized collaboration between departments like public security and transportation.
4. Customer Benefits: From “Seeing” to “Understanding” with Intelligent Leap
- Resource Efficiency Revolution: The utilization rate of the virtualized resource pool increased by 50%, significantly reducing hardware investment costs.
- AI Analysis Efficiency Breakthrough: GPU scheduling efficiency improved by 30%, with the time for processing critical area surveillance tasks reduced to seconds.
- Standardized Replicable Model: The platform architecture became a nationwide benchmark for CCTV cloud transformation, with rapid replicability.
In August 2024, ZStack officially launched its Globalization 2.0 strategy, building on the success of its Global 1.0 phase in serving over 30 countries and regions worldwide. The Asia-Pacific region is positioned as the cornerstone of this strategy, deepening international partnerships, focusing on resources and investment, and increasing regional penetration. ZStack aims to become a major global intelligent computing platform provider. This project not only validates ZStack Cloud’s global service capabilities in multilingual support (9 languages), heterogeneous computing management, and high-security compliance but also highlights its unwavering commitment to supporting the intelligent upgrade of critical infrastructure worldwide.
