The direction against the wind is better for flying.
Preface: There are four parts that need to be considered for the monitoring project, which are the front-end system, the transmission system, the storage system and the display system. As for the transmission system, we have mentioned it many times before. The storage system is very important in the monitoring project and many people often discuss it. Therefore, today we will take a look at the differences between NVR, IP-SAN and CVR, which are three commonly used storage methods.
Ⅰ. IP-SAN storage mode
IP-SAN is a SAN storage framework based on the IP Ethernet network. It uses the iSCSI protocol to transmit data and store them directly on the IP network. The iSCSI protocol transmits SCSI command packets in TCP/IP packets, which is SCSI over TCP/IP. IP-SAN framework uses Ethernet networks, Ethernet cards, and iSCSI storage devices. IP-SAN can divide a storage device into one or more volumes and export them to the front-end application client, and the client computer can create a new file system (format) for these imported volumes. The client computer accesses these volumes through device-level block access. IP-SAN divides data into multiple data blocks to write or read disks in parallel. The characteristics of block-level access determine the high I/O performance and low transmission latency for iSCSI data access. IP-SAN inherits the advantages of IP networks such as openness, high performance, high reliability, easy management, strong scalability, and strong adaptability. It has a flexible storage mode, which achieves a seamless connection between the storage network and the application network and provides excellent remote data replication and disaster recovery features.
1. IP SAN storage framework
Video surveillance images should be stored with a high degree of reliability, and it is not allowed that there is no video or cannot be played back when the video needs to be played back at a certain time and place. The storage system should be designed and configured in the following ways to improve system reliability:
(1) Improving the redundancy technology of the storage server to ensure the uninterrupted work of storage management.
(2) The storage disk array adopts RAID technology and various hardware redundancy to ensure the high reliability of the system, and a disk failure will not affect the continuous data access.
(3) The online disk update mechanism uses a hot spare disk to automatically replace faults.
2. IPSAN storage features
(1) It has the advantage of high bandwidth "block" level data transfer.
(2) It is based on TCP/IP. The IP network technology is mature, and it combines advantages of TCP/IP, such as reliable transmission and routable, which reduces the complexity of configuration, maintenance and management.
(3) The iSCSI storage network can be deployed through Ethernet, which is easy to deploy at a low cost.
(4) It is easy to expand. When the storage space needs to be increased, you only need to add storage devices to have high scalability.
(5) It has easy data migration and remote mirroring. It has no distance limit as long as the network bandwidth supports it, which can better support backup and remote disaster recovery.
Ⅱ. CVR storage mode
1. The CVR storage mode supports video streaming directly into the storage device through the encoder, which saves costs for the storage server and avoids the single point of failure and performance bottleneck of the server. The unique data structure ensures high stability and high performance of the monitoring service.
2. Features of CVR storage mode
(1) The front-end direct-write unified management CVR storage mode can support video streaming directly to the storage device through the encoder without storage servers. The centralized management platform enables the centralized management and status monitoring of storage devices on multiple networks. The centralized configuration and management of storage devices in the service system can be achieved, and the working status of storage device units and their corresponding manageable device components, operating protocols and RAID groups can be monitored in real-time. The centralized alarm management supports regular status inspection of storage devices to monitor and display the status of the system, array operation, ISCSI, HTTP, CIFS, SCSI channels in a centralized manner in real-time.
(2) Simplify the network structure: In CVR storage mode, front-end coding devices are directly written to storage devices without servers. A simple network structure can relieve the network pressure between the storage server and the storage device in the original design.
(3) Unique streaming media file system protection technology: The CVR storage mode adopts unique VSPP-video stream pre-protection technology to completely solve problems such as monitoring service stop, data read-only or loss caused by file system instability or even file system damage caused by power failure and network disconnection. The CVR storage mode adopts its unique data block management structure and fault tolerance mechanism to separate itself from the server file system and avoid the impact of its file system damage on the monitoring business system, and provides a more stable and efficient management method.
(4) Efficient disk fragmentation immunity technology: CVR storage mode adopts the combination of disk pre-allocation and delayed allocation technology, which can find free space area and use it to store new data to maximize system performance and avoid disk fragmentation. At the same time, it combines the efficient defragmentation program to defragment the disk when the system is idle and improve the system performance.
(5) System openness: The CVR storage mode supports RTSP/SIP/ONVIF/GB/T28181 protocols, and is compatible with front-end coding devices that support standard protocols for direct access.
(6) Number and capacity of video volumes: The surveillance storage system proposes the concept of video volumes, and it uses video volumes as a unit to provide external space. Each video volume supports up to 1024 logical volumes, so the single-capacity space can be up to thousands of PB. 7. Efficient positioning and retrieval in seconds.
(7) The CVR storage mode system has a unique fast indexing method "device number->corresponding data volume>offset”. The capacity of the data volume can be as high as TB or PB level, which minimizes the retrieval depth and improves the retrieval efficiency.
3. NVR storage
NVR refers to Network Video Recorder. The main function of the NVR is to receive the digital video stream transmitted by the IPC (network camera) device through the network, and to store and manage it to achieve the distributed framework advantages brought by the network. Simply speaking, you can watch, browse, playback, manage, and store multiple network cameras through NVR at the same time, which can get rid of computer hardware and there is no tediousness of installing software.
Ⅲ. Features of NVR storage mode
NVR adopts an embedded system for framework, and it inherits the excellent advantages of Linux or embedded system in system operation and maintenance stability, which has better stability and system operation cost performance. It can use a simple framework and low equipment cost to create a monitoring framework comparable to professional servers. In terms of video channel limit, when the number of video channels supported by a group of NVRs is limited, the framework of multiple NVRs can be adopted to achieve the system service energy required by the network video monitoring system.
Ⅳ. System comparison
1. Comparison of disk array and NVR
(1) Storage location
Most NVRs are front-end deployments and edge storage. Disk arrays belong to the centralized storage mode in the central equipment room.
(2) Data protection
Many NVRs on the market claim to support RAID technology, but they are limited by the performance of embedded chips and resulting in poor RAID performance. RAID technology is the core of the disk array, which can effectively protect data integrity. RAID drive failure can be continuously optimized through the system layer and driver layer to better meet the monitoring market.
(3) High density
High-density disks are centrally stored in a single column, which effectively saves equipment room space. The same storage space requires a smaller number of disk arrays, which reduces the maintenance workload of administrators and the number of faulty nodes.
(4) High performance
NVR adopts an embedded hardware framework, which is affected by the processing capability of embedded chips. It supports up to 128 channels of D1 video input. However, when 3T and 4T large-capacity hard disks are introduced, NVR performance is often failed to meet. The disk array uses a high-performance hardware framework to provide higher video access performance and 288 channels of D1 video access.
NVR adopts embedded hardware design, and the hardware and software specifications of the device have a small space for upgrading. The disk array has flexible software and hardware upgrades, which supports hardware expansion and third-party software integration.
2. Comparison of disk array and CVR
The CVR front-end uses streaming media protocol to write data directly to the storage without network mounting, and the CVR takes the initiative to fetch streams. Each front-end is mounted with independent iSCSI storage space in IPSAN mode, which is uniformly allocated by the management server.
CVR does not rely on any intermediate links while IPSAN relies on intermediate management servers.
CVR supports multi-protocol access such as RTSP/ONVIF/SDK/SIP, and supports GB/T28181 as well as custom development. IPSAN is relatively closed and requires front-end support for iSCSI.
In the case of a poor CVR network, you only need to establish a session connection, and the reconnection process takes a short time. IPSAN is easy causes connection interruption due to poor network. Therefore, the front-end and storage devices must constantly be trying to remount the connecting, which is time-consuming and easy to lose video.
(4) Video application optimization
CVR solves the problem of overwriting fragmentation and file system instability after power failure. It can also be used for PB-level video volume management, integrated video management and frame loss/recording detection. IPSAN is based on multi-LUN volume management, which affects performance and cannot avoid problems such as traditional file fragmentation and power-off file system instability.