Background

An operator’s most direct physical marketing channel is the business hall. Business halls play a unique role in presenting corporate image, promoting service brands, and raising perceived customer quality. To keep pace with mobile-communications development and to keep business halls orderly and safe, Hubei Mobile needed a cloud monitoring system that integrates brand promotion, service content, promotions, staff training, incident handling, and business-hall security, providing users with a better information platform.

Requirements

1. Based on a total of 8,000 video storage channels, a 1:1 mix of 1080P/720P and 480P resolution, and a retention model of 1 day : 7 days : 15 days : 30 days at 1:79:2:18, the system must meet the corresponding video storage capacity.

2. The existing monitoring architecture cannot meet later big-data analysis requirements.

The original monitoring system installed multiple cameras in each business hall. Cameras stored monitoring data via DVR or NVR. Video analysis required additional dedicated servers, another copy of the video stream, and then analysis. The storage platform was typically Hadoop. Because HDFS generally uses three replicas, the video stream from each camera had to be stored four times before analysis results were produced. This increased investment and was very inefficient.

3. Operations issues

Business outlets are widely scattered, and video monitoring equipment cannot be managed and maintained in a unified way. Outlets generally have no dedicated operations staff. Any problem at any outlet requires the operations department to handle it, so resolution is slow and operations pressure is high.

4. Data-reliability issues

NVRs or DVRs at outlets are typically 8-bay or 4-bay appliances. Data reliability depends on RAID. Once a disk fails, traditional RAID rebuild is slow, and rebuild pressure may cause video-stream frame loss. NVR disks are mostly surveillance-grade disks with a relatively high failure rate, further reducing data safety.

Solution

The upper-layer video-monitoring software is ZTE Liwei. Sugon recommended the technically strong cloud-storage solution ParaStor video-surveillance storage to support 8,000 front-end cameras. The monitoring platform’s storage module is reused with ParaStor data nodes, effectively reducing front-end servers. Each data node is configured with two 10G ports for the internal data network and management network, and four Gigabit ports for front-end business access.Through the video-monitoring platform, camera video streams from business outlets are ingested into Sugon cloud storage at the back end for centralized storage. After multiple rounds of testing and technical review, a converged cloud-storage design was adopted, deploying the video-monitoring platform’s storage-management module on the cloud storage Ostor nodes and reducing the number of storage-management servers purchased.