Video archiving systems: distinctive solutions and trends.
Despite the increasingly widespread introduction of digital technologies, the question of choosing an analog, hybrid or fully IP system is still open for the customer. Now the type of system is determined by the selected cameras. But whichever system the user chooses, video information will be stored digitally and on hard drives (HDD). Therefore, the question of trends in the field of video storage will be reduced to one thing: how to record from a camera directly to a hard drive.
As for video storage systems, we have suddenly abandoned analog VCRs and video cassettes (although the latter are widely used in many systems) and switched to storing information in digital form. What advantages have been gained by switching from VCRs? VCRs with analog storage do not allow for prompt search through the archive. In addition, further processing of analog information is practically impossible, since the VCR recording has lower quality than the original signal.
Despite the ever-wider introduction of digital technologies, the question of choosing an analog, hybrid or fully IP system is still open for the customer. Currently, the type of system is determined by the cameras selected. But no matter which system the user chooses, video information will be stored digitally and on hard disks (HDD). Therefore, the question of trends in the field of video storage will be reduced to one: how to record from a camera directly to a hard disk.
As for video storage systems, we have suddenly abandoned analog VCRs and video cassettes (although the latter are widely used in many systems) and switched to storing information in digital form. What advantages have been gained by switching from VCRs? VCRs with analog storage do not allow for prompt search through the archive. In addition, further processing of analog information is practically impossible, since the VCR recording has lower quality than the original signal.
Compared to a video cassette, and other storage devices (flash card, CD or DVD), a hard drive has a significantly higher data transfer rate and greater capacity, allowing you to record video with a quality higher than S-VHS, using the appropriate analog video compression algorithms. Hard drives with a capacity of 750 GB are now widely used, which solves the recording duration problem that existed just a couple of years ago.
This article attempts to trace trends in the field of information storage. For clarity, we will consider a distributed (up to 2000 cameras) video surveillance system. We will trace possible options for deploying the system and options for storing the received video information.
The task of video surveillance can be divided into three components. The first is the eyes of the video surveillance system — cameras. At the moment, you can choose analog, IP (704 x 576 pixels) or megapixel cameras, which basically determines the future system. The image obtained by the cameras must be delivered to the operator to make prompt decisions depending on the observed scene. In addition, we must ensure the safety of video materials. The ideal video surveillance system is the sum of three independent components: receiving an image, viewing it and storing it.
Video recorders
When using analog cameras, it is necessary to use devices for encoding an analog signal into a digital one in order to ultimately store the received information on hard drives.
A digital video recorder (DVR) is a device to which coaxial cables from analog cameras are connected. Typically, such a device has internal or external hard drives for storage. There are a huge number of DVRs, such as PC-Based or Stand Alone, with Windows or Linux operating systems, with installed or expandable capture cards, with a different number of channels (4, 8, 16, 24, 30). However, their operating principle remains the same — signal digitization and storage. Real-time images can be obtained directly from the DVR or DVRs can be combined into an Ethernet network for remote viewing by the operator. An obvious disadvantage of such a system is the likelihood of DVR failure. Moreover, the operator loses the ability to monitor from cameras, and the DVR, due to its failure, stops not only recording, but also accessing the previously recorded archive. For our proposed system of 2000 cameras, 125 16-channel DVRs will be required.
Network video recorders
In the example described above, DVRs have a limit for storing video information. The internal storage capacity of a DVR is about 2 TB, which must be divided between all cameras. Many DVRs have the ability to increase disk space by connecting external storage (USB, SCSI) and reach tens of terabytes. However, the DVR still has to be a potential point of failure, and in addition to this, it digitizes the signal, records and transmits video for viewing (multitasking).
Therefore, it is logical to use professional video storage devices. There are such solutions on the IT market. Why reinvent the wheel? We will separate the storage device from the video recorder. This device is called a network video recorder or Network Video Recorder (NVR). NVR is software installed on a server connected to an Ethernet network. Disk space is created on the server (internal SATA disks or external SCSI arrays). SCSI (Small Computer System Interface) disk arrays are a basket in which hard disks are installed, forming a high-speed and reliable storage system. Currently, such arrays have an average of 16 trays for installing disks with a capacity of 500 GB, 750 GB or 1 TB. Due to the growing requirements for the size of disk space and reliability of data storage, RAID arrays (Redundant Arrays of Independent Disks) are used. RAID combines several hard drives, if one of the drives fails, the data will not be lost and the failed drive can be replaced during operation.
Thus, the video recorder is divided into two independent systems: storage and image retrieval, united by a local network. You will need devices capable of receiving a digital stream and transmitting it over the network to the network video recorder. You can use IP cameras that allow you to receive a digital stream (MPEG4), which is sent to the NVR. To digitize analog cameras, video coders are used for 1, 2, 4, 8, 12 or 16 channels. Unlike DVRs, video coders only perform the function of encoding an analog image and transmitting it to the network for viewing by the operator and recording on the NVR. As a rule, therefore, manufacturers of surveillance systems provide IP cameras and coders with the ability to create two independent streams. One stream for viewing is 25 fps with a resolution of CIF, and the other from 4 to 12 fps, 4CIF. Although it is possible to configure both streams for maximum performance (4CIF@25ips, ~ 3Mbit/sec).
Modern network video recorders are capable of recording up to 300 cameras. For a system of 2000 cameras, depending on the quality of encoding, about 50 servers will be required. However, although such devices are reliable, they require appropriate maintenance. The network recorder runs on the Windows Server 2003 operating system, which requires updates, and like any computer, requires anti-virus protection. Often, a fail-safe Failover NVR is used for backup, which is capable of accepting digital streams in the event of a failure of the main NVR. Thus, the next stage in the development of storage systems is the exclusion of the server.
Direct recording
The ideal IP video surveillance system is the sum of three components (image acquisition, viewing and storage) connected by an Ethernet network. With the advent of Fibre Channel (FC) and SAN built on it, the storage world has placed its bets on network access to storage systems; the future belongs to data storage networks. For several years, the FC interface remained the only standard for their construction, but today many people understand that it is time for a change. A very interesting storage system is iSCSI (Internet Small Computer System Interface). iSCSI is a protocol based on TCP/IP and is designed to establish interaction and control of data storage systems, servers and clients. Without going into technical details, I will note that on the video surveillance systems market, the device is a basket for installing hard drives, reminiscent of SCSI, but on the back of such a basket, instead of the usual SCSI interface, there are RJ45 connectors for connecting the array to a local network.
,
As shown above, streams from IP cameras or IP encoders are transmitted over a local network. The iSCSI disk array receives a digital stream directly from an IP device over the network and places it on the array's hard drives. A computer is only needed at the system configuration stage. The essence of the configuration is that the disk is divided into logical sections (LUN), each LUN is assigned an IP address of the camera or encoder. However, if it is necessary to increase the capacity allocated for the camera, problems arise. When reformatting, data is lost.
This option has proven itself well for small hybrid systems, for example a multi-channel encoder (for 16 cameras and having two gigabit ports) is connected directly to an iSCSI disk array. Thus, the local network is not loaded with a video stream directed to recording.
As we know, for example, digital video in MPEG4 format has a variable bit rate, which means different storage volumes for images from a camera observing a dynamic scene (entrance to an office, especially during rush hours) and a camera installed in a quiet backyard or basement. Therefore, it is impossible to accurately calculate the space needed to store video for, say, 7 days. We either create sections that are too large (the customer overpays) or too small (the customer is angry: instead of 7 days, some cameras only record 5!).
Therefore, some tool for disk space optimization is needed. For example, in our project we will use 80 iSCSI disk arrays of 6 TB each for 2000 cameras, which makes it practically impossible to configure manually. Some management software is needed (let's call it Video Recording Manager, VRM), in which you simply specify IP cameras/encoders and iSCSI disk arrays located in the system.
VRM introduces the concept of a virtualized storage layer. This abstract layer allows all individual disk arrays in a system to be managed as a single virtual shared storage pool that is intelligently allocated as needed. VRM eliminates the need for network video recorders and the associated server hardware, operating systems, antivirus software, and the constant software patches and updates that these systems require. With this new technology, installation, operation, and maintenance of the system become much simpler and more cost-effective. The number of arrays can be increased at any time, and the quality and duration of recording can also be changed depending on current needs.
Currently, iSCSI technology operates at speeds of up to 1 Gbps. In the future, 10 Gbps technology is expected. However, iSCSI currently has a competitor in the form of Fiber Channel with a speed of 4 Gbps. IP equipment manufacturers are preparing for the transition to Fiber Channel, with some video encoder models featuring an SFP slot that can be used for both optical Ethernet and Fiber Channel storage.
Technologies and speeds change, but one thing remains constant – the hard drive.