Application of Cloud Monitoring in the China Unicom International Multimedia Transmission Platform

China Unicom

1. Project Background

In 2015, China Unicom's management proposed the strategy of "implementing a focused strategy with innovative and collaborative development," positioning TV video services as a strategic foundational business for China Unicom. Providing customers with low-latency, high-speed, and secure transmission channels can effectively enhance China Unicom's competitiveness in the video services market. In accordance with this strategic plan, China Unicom decided to build a multimedia transmission platform. Phase one covers key domestic cities and selected overseas POP sites, deploying multimedia transmission platform equipment along with media gateways, encoders/decoders, and other devices. Using dedicated lines or public networks, it addresses high-quality real-time video and audio transport, aggregation, and distribution between overseas POP sites and domestic locations, as well as among key domestic cities, serving important customers in broadcast / broadcasting, online video providers, and other sectors. Phase two extends to key overseas regions, enabling low-latency, highly reliable real-time audio and video transmission between domestic and international locations. The video media network is centered on Beijing as the media hub, connecting domestic and international POP sites to form the China Unicom video media network, covering domestic and international media clients and providing global media content delivery services.

The multimedia transmission platform is a value-added service built on OTN networks combined with fiber optic, satellite, dedicated line, and internet transmission networks. It primarily transmits video, data, and other content, serving domestic and international media clients, internet video companies, and video rights companies, delivering significant economic benefits. It provides professional multimedia video transport services for enterprise customers, building a global video media network that meets customers' needs for flexible networking and broad video content delivery through the multimedia transmission platform.

II. Challenges

1. Cross-regional coverage

Geographically, the entire transmission network spans multiple cities across China from east to west and north to south, as well as overseas locations including Hong Kong, Tokyo, London, and Los Angeles, crossing continents and oceans. Currently, the multimedia transmission platform has nodes in 9 domestic cities: Beijing, Shanghai, Guangzhou, Tianjin, Wuhan, Shenyang, Xi'an, Chengdu, and Fuzhou, as well as 4 overseas cities: Hong Kong, Tokyo, London, and Los Angeles. In phase three, the following nodes will be added: Changsha station, Zhejiang station, Urumqi station, satellite station, CCTV station, and mobile station.

The transmission network spans multiple cities domestically and internationally, crossing continents and oceans
The vast geographic span and long distances pose inherent challenges for building a unified network management system.

2. Users are distributed across different departments and branch offices, requiring role-based access control with domain separation to support larger organizational structures

System scale, complexity, and security assurance: when the system scale and number of users reach a certain level, access security considerations require assigning different permission levels to simplify operations and management. Otherwise, flat or identical account permissions would cause resource access conflicts and management confusion. For cross-regional systems, when controlled devices and users are located in different places, permissions must be divided by region so that local users can only access local device resources.

3. Large system scale

This is reflected in the large number of devices, segmented networks, and numerous monitoring points, as well as signal diversity including format conversion.

System network overview (Phase 1)

4. Complexity

Massive numbers of devices, enormous signal volumes, diverse signal formats, work-order-driven alarm filtering arising from business process automation, and the demand for intelligent linkage / interlock all increase overall system complexity.

4. Linkage / interlock with booking and automated transmission business systems is required

This system must interconnect with China Unicom's existing systems, including booking, routing, signal scheduling, transmission, and business reporting. The transmission platform is not an isolated island but an integral part of a unified ecosystem with the existing systems. System-to-system interaction means these web applications should use well-designed API calls for cross-system access, enabling information flow and remote control relay while ensuring system access security.

III. System Components

Transmission Platform Node Architecture

IV. Monitoring System Functions

Cloud Monitoring is a comprehensive monitoring platform designed for cloud and IP-based environments. Built on a microservice architecture, it supports hierarchical monitoring through a proxy mechanism, providing unlimited scalability.

The platform provides comprehensive monitoring with automation and linkage / interlock, agile and responsive. It supports big-data-based statistics and prediction, pattern extraction and trend analysis, report generation, operations optimization and recommendations. One-click cloud deployment with B/S architecture enables quick browser access.

Cloud Monitoring home page

1. Software functions

(1) Intuitive and vivid real-time display of system status in IP environment;

Real-time parameter display

Real-time alarm display

(2) IP-based audio/video stream monitoring;

Real-time alarm list display

(3) Signal adjustment, scheduling, and distribution;

(4) Intelligent discovery of devices in IP environment;

(5) Complete device monitoring capability;

Parameter configuration and adjustment

(6) Complete signal detection capability;

Signal link monitoring

(7) Complete network monitoring capability;

(8) Data collection and data analysis capabilities, including aggregation, storage, modeling, processing, and operational recommendations;

Data processing and analysis platform

• Enterprise-grade post-event data analysis and processing platform;
  
• Suitable for collection, aggregation, storage, and online query processing of massive data from various sources;
  
• Leverages NoSQL databases such as Redis and MongoDB for superior performance and availability;
  
• Supports clustering and horizontal scaling with parallel processing;
  
• Online report generation;
  
• B/S architecture with multi-user concurrent access support;
  
• Simple, easy to maintain, easy to deploy;
  
• Supports high-speed batch data processing;
  
• Provides a GUI for interactive data export;
  
• Supports export in multiple file formats including XML, JSON, MS Excel, and CSV;
• Supports multi-source data aggregation to generate unified views and reports;

Big data analysis platform

NoSQL database replica sets and sharded clusters provide superior performance and availability

Metric parameter statistics and indicators

Various report outputs

(9) Role-based access control with domain separation

Role-based access control with domain separation covers user management, department and organization management, permission management, and view partitioning. Permission management is the focus, as multi-user, multi-site, concurrent access requires robust authorization and permission checking. The system provides mechanisms including login authentication, department management, and permission control for device read/write, signal scheduling, and link status display. Domain separation primarily authorizes users based on their department (e.g., regional branch offices) to access only specific topology diagrams and device sets. For example, users at a remote branch office can only access devices and corresponding topology diagrams and signal flow diagrams in their local facility. This is built upon the access control mechanisms described above. User management assigns different permissions based on user roles (super administrator, guest, operator, etc.) and departments. Permissions can also be divided by account role or by department affiliation. Rigorous permission management ensures that unassigned access rights will not allow resource access.

Authorize users by department

Strict authorization verification

1. Non-functional requirements

(1) Ease of use;

Virtual desktop with customizable home page to simplify daily use

(1) Customizable user interface and experience;

Alarm suppression and filter masking

Combined display, dashboard real-time parameters

V. System Features

1. B/S architecture, enabling anytime-anywhere system access;

Provides diverse clients including browsers for anytime-anywhere system access

2. Comprehensive solution;

3. High degree of automation;

Supports linkage / interlock with booking, transmission, and other business systems for work-order-based alarm filtering. Provides event response mechanisms for linkage / interlock, follow, and other automation modes. Offers standard Restful APIs on which workflow automation systems can be built.

4. Excellent elastic scalability;

Supports secondary proxies, device polling proxies, and clusters. Horizontal scaling through additional proxy instances supports greater system loads.

Device driver proxies are the key to handling diverse control needs and massive device volumes

The device driver proxy is based on a microservice architecture, adapting to different interface types, protocol specifications, device types, and access media. Key features are as follows:

• a) Microservice modules conform to unified standards and can be flexibly plugged and combined to adapt to diverse device types and communication methods;
• b) Provides Restful APIs that offer a unified interface to the monitoring center for reporting information and executing commands;
• c) Supports caching and message bus mechanisms, leveraging read/write separation and cluster load balancing to improve performance;
• d) Supports the proxy pool concept, enabling horizontal linear scaling through multi-instance deployment to support more devices and wider deployment scope;
• e) Supports distributed deployment across machines, network segments, and regions. For example, driver proxies can be deployed close to controlled devices in the same facility or even the same rack for intensive device access, or within the same network segment as controlled devices for enhanced access control and security.

• 
  Multi-instance dynamic service deployment is the key to horizontal performance scaling

The microservice module Supervisor is used to supervise and perform scheduling for the device driver proxies and related microservice modules. As a core infrastructure component, the Supervisor closely monitors the above microservice modules, especially their health status. Specifically:

• a) Performs health checks and liveness checks on supervised modules;
• b) When individual microservice modules malfunction or crash unexpectedly, the Supervisor automatically detects and restarts them;
• c) Load balancing, especially dynamic load balancing. Operators can manually deploy, scale instances, and start or stop services via the GUI based on load and health status. Alternatively, automatic mode can be enabled with specific rules to automatically create, deploy, and remove instances based on load and module health.
• d) Supports one-click primary/standby switchover for business services and proxies;
• e) Provides Restful APIs that publish the health status of proxy modules and other microservice modules. Based on these microservice module health statuses, linkage / interlock and intervention actions can be performed, such as SMS notifications, email notifications, and dynamic load switchover.

• 
  Device driver proxy list

5. Microservice architecture for rapid adaptation to customized business workflows;

The Microservices Architecture Pattern constructs large, complex, long-running applications as a set of coordinated services, each of which can be easily improved locally. Since each service can be independently scaled on the x-axis and z-axis, and each service can be deployed on appropriate hardware servers based on its own requirements, horizontal performance scaling is achieved through multi-instance deployment.

6. Provides multi-level, multi-source data collection and aggregation capabilities, driving continuous operations optimization through big data analysis;

Capable of collecting device configuration parameters, operational metrics, device status, signal quality, network traffic, send/receive activity, server operational metrics, user access activity, facility environment data, and more;

Real-Time Data Processing platform

Supports alarm analysis based on real-time parameters, such as unexpected configuration changes, threshold violations, target disappearance alerts, and unexpected target appearance alerts. Also supports multi-parameter joint analysis with rule-based formulas to define parameter constraints and correlations for logical inference and trend analysis.

7. Combined with data analysis and processing, provides operational reports from multiple perspectives to enhance insight into system performance;

Daily operations trend analysis

Parameter value time-history curves

7. Standard API interfaces, Restful API

(1) Standard interface using HTTP protocol + JSON, human-readable, simplifying firewall and port configuration;

(2) API coverage includes resource access, capability access, function access, and common services;

(3) Provides comprehensive online documentation that stays synchronized with the APIs in real time;

(4) Enables building workflow automation systems such as automated ingest, Task Execution, and follow-on linkage / interlock;

V. Project Timeline and Results

Cloud Monitoring completed system debugging and testing from mid to late 2017 and was delivered for use in the first half of 2018, running stably ever since. During this period, the Cloud Monitoring software, together with the multimedia transmission platform, supported live production for major events including the Winter Olympics, the Asian Games, and the World Cup. Its powerful monitoring capabilities, intelligent user experience, intuitive interface, comprehensive data collection and analysis, and versatile report outputs have earned client recognition.

China Unicom Video Intelligent Premium Network Operations Service Center

During this period, Phase 2 expansion added overseas nodes in 4 cities -- Hong Kong, Tokyo, London, and Los Angeles -- on top of the existing domestic cities. In 2019, Phase 3 expansion added domestic nodes: Changsha station, Zhejiang station, Urumqi station, satellite station, CCTV station, and mobile station. As the multimedia transmission platform expanded in phases, Cloud Monitoring demonstrated excellent scalability, particularly its horizontal scaling capability. The application of core frameworks such as the microservice architecture and device proxies ensured seamless system evolution and elastic scaling, while leaving room for further future expansion.

As one of the most important auxiliary systems, Cloud Monitoring safeguards secure and high-quality broadcasting for the China Unicom International multimedia transmission platform.