It has close relationship between the development of video surveillance area and video surveillance itself. Video surveillance has come across three stages: analog, digital and Internet. So as the storage of it.
    The collection of video and storage of video surveillance
    Video surveillance system normally possess the character of varied surveillance positions, large numbers of cameras and long time of monitoring etc, so the reading and writing methods of the storage devices distinguish from other systems. First, digital video encoder or video server write data into storage device in streaming media mode, and save the images and pictures returned from control spots, playback stations read the stored video files on streaming media method. It’s totally different from the small piece of data and documental read and write method used by ordinary database system or file server system, so the requirements of parameters in video surveillance system are quite different from others. Second, large video surveillance system often has 2000-3000 monitoring cameras or more, totally need 500Mb/s～750Mb/s bandwidth.
    In the process of video collection, video file format generally will not change, and keep CBR rate. Video image frame rate is usually in 15 frame/s～25frame/s. That is to say, except that the storage device configures a large cache, it must guarantee 500Mb/s～750Mb/s bandwidth in 1/25s in read and write operation of the storage, or will loss frames when collecting and playing back video. It shows that the storage of video surveillance system need large and constant bandwidth. Third, the reading and writing data operation last a long time. As the cameras normally work 7 × 24 hours, even the streaming media file use segmented storage, the duration of the writing operation is also possible for up to 2 to 6 hours, later playback also requires the same time. Thus requiring large storage with superior long time working ability and stability. Forth, video surveillance system normally has the characters of  large numbers of cameras, long time video storage etc, so a huge demand for storage capacity, and with the image storage time increases, the demand for storage capacity is linear, explosive growth. Therefore, video surveillance systems must support large capacity and highly expandability storage to meet the long time and high-capacity video storage needs. Fifth, Digital video encoder provides a standard IP interface, or iSCSI interface, through ADSL, metropolitan area network or private network to transmit TCP / IP protocol packet, or iSCSI protocol packet to return data to the rear control center. If the storage device has the same interface with digital video encoder, the data may be written directly to storage by encoder, thus greatly reducing the data transmission links of monitoring system, improve storage efficiency.
    From the gradually development trend of video surveillance to networking, network storage will play an increasingly important role in the future of video surveillance systems. Network video monitoring system take storage into consideration from two aspects: First, the deploy and access of the network, taking into account the reasonable allocation of flow and storage room, the deployment location of storage devices throughout the system will tend to distributing and front-ending. And from the perspective of easy unified access for users, all the storage space should be transparent to them, the user can access for all storage resources only through a single interface. Second, storage device itself for networking, network video surveillance is very suitable for large-scale network monitoring applications. In order to meet the needs of large-capacity compact storage, storage device itself should use  advanced network storage technology to ensure flexible space expansion.
    Front-end storage
    The so-called front-end storage, is that storage unit is built into the front-end equipment (such as network video encoders or IP cameras) of  the network video monitoring system. And implement monitoring image’s local record and save by the front-end equipment directly.
Front-end storage has several advantages: First, through the deployment of distributed storage, reduce the capacity pressure bringing by centralized storage; Second, effectively alleviate the network flow pressure caused by centralized storage; Third, avoid image lost in the network failure caused by centralized storage.
    For the front-end storage, usually because a single front-end coding equipment bring few monitoring points and short storage time, storage capacity demand is not high. IP cameras generally use CF card or SD card, video server with built-in HD. This is basically no difference comparing with the previous stand-alone store.
But what is essentially different is that, in order to ensure flexibility and easy user access, network video surveillance systems not only need to provide stand-alone point to point access in all front-end storage, but also provides all the content shared through a unified centralized interface. To this end, network video monitoring system manage unified management and scheduling of all front-end storage by the central service platform, and to achieve the storage space and storage contents networking. In this way, users can either log in directly to a single front-end device or unified log in central service platform for recording data on-demand or playback.
    Central storage
    In the network video surveillance system, the deployment pay more attention to central storage. Front-end equipment collect images from monitoring point and encoded into a digital surveillance stream, then transmitted through the network to the central service platform, at last, stream will be distributed to network record unit and stored centrally.
    In many large-scale network video surveillance applications, also can use center storage method of multi-level distribution, i.e., sub-center storage, so that it can reduce pressure of storage capacity and network traffic brought by centralized storage in a center point and greatly enhance system reliability.
    Using center / sub-center storage, it has obvious advantages in the following aspects: First, for users, more convenient to retrieve and call the video resources; Second, easier to guarantee the completeness of storage contents, and will not loss important content caused by theft or damage of the front-end equipment; Third, can saving resources through reasonable resource scheduling, and allocating storage space for the front-end equipment according to need; Forth, beneficial for developing diversified storage strategy to meet individual requirements of users; Fifth, easy to maintain, convenient for concentrated check and timely troubleshooting.
    For applications to fewer monitoring point, short storage time requirements, the center/ sub-center storage can use the relatively simple deployment methods of hard disk insertion ro server or external disk cabinet accession, known as DAS (Direct Accessing Storage), and similar to standalone storage. While the advantages of network video surveillance is widely accepted, it begins to appear more and more large-scale and super video surveillance system now, such as social security monitoring system in "Safety City" project, and two carrier class video surveillance systems of China Telecom and China Netcom in processing "Global Eye" and "Wide Vision" projects. These kinds of monitoring systems are all facing the needs of front-end equipment accession and large capacity centralized storage. Previous stand-alone storage can not meet the application needs that flexibly expand systems capacity, it must use  more advanced network storage devices and storage technology, among which, SAN, NAS and iSCSI are the typical representatives.
    SAN (Storage Area Network) originated in the late nineties of last century, different from DAS, SAN based on Fibre Channel technology. Servers connect storage arrays via Fibre Channel switches, form a private area network for data storage. SAN uses a network-oriented storage structure, separation of data processing and data storage, with characters of easy expansion storage space, flexible addressing, long-distance data transmission, high I/O performance, and high utilization of storage devices, is a full new storage system.
    Different from SAN which specially based on fibre channel protocol, NAS (network accessing storage) achieved interconnect between servers and storage arrays basing on IP network,           make use of TCP / IP protocol for communication, transmit data in file-class I/O method. By contrast, the installation, commissioning, usage and management of NAS devices are simpler, the deployment cost is relatively low.
    iSCSI is a new IETF standard protocol, that is a transmission standards transform SCSI block data into a packet network through the IP network. Like NAS, it transmits data through the IP network, but it uses different stored way, and the same Block Protocol with SAN. Thus, iSCSI bring user the value of: First, iSCSI enables SCSI packets to be transmitted in Ethernet, so that make SAN free from expensive fiber networks, can realize original function by IP networks, which reduces management complexity and cost; Second, because of the complexity of user applications, it will often deploy two storage network, SAN and NAS, and iSCSI can integrate the two.
    These features of iSCSI is a perfect fit with current developing status and directions of video surveillance, especially in the field of carrier class video surveillance which needs large-scale storage, high investment. Basing on the current mature IP network center/sub-centers storage systems building, iSCSI Technology is undoubtedly a good reference.
Application Storage
    The often used storage network structures of video surveillance systems are DAS, NAS, SAN and IP-SAN, and the often used storage devices types are SCSI disk array, FC and iSCSI devices. These devices are in the position of mature technology, simple structure and easy installation. However, these devices have a common disadvantage that storage devices in video surveillance systems only can play a simple data storage function, can not replace the server to run specific application software, video surveillance systems still need a lot of special servers.
    In fact, with the rapid development of chip technology, the calculating and processing speed of CPU and dedicated ASIC chips improved by times, but the overall price of storage continues to decrease. In equipment selection and purchasing, the users will generally choose a higher level of storage devices in considerations of equipment upgrade. In addition, when investment is available, the impact of the price will make relatively smaller difference to users. Thus in many application systems, the performance of storage devices brought by users are usually higher than the real needed business systems, even beyond several times.
In some small video surveillance system, the performance bottlenecks of the system often appear in the sides of video server rather than the storage devices, storage devices often can provide twice or three times bandwidth more than the actual total bandwidth of video surveillance, IOPS is also 3 to 5 times than the actual demand. Storage device controllers are in a semi-load working condition for a long time, to cause to waste a large number of high-performance resources. To ensure high-availability, high efficiency, stability and security, the controller of storage devices are generally specially designed, using a dedicated processor and cache, its technical parameters and performance are higher than normal PC server. In order to effectively utilize the surplus storage resources, application software with special features can be embedded in the storage device controllers, storage devices not only can provide data storage server, but also can provide a number of software application servers, this kind of storage device is called application storage.