user guide
utilities
Go to Website For Partners
Product Line-up

home > learning resource > white papers > hard disk drives for video surveillance

Hard Disk Drives for Video Surveillance

Video Surveillance Market Overview
Video Surveliiance can be segmented into several markets -
  • High end - casino, government, campus, banking, airport
  • Mid-range - office, health care facility, warehouse, condo, hotel
  • Low end - small business, residential
  • Embedded - ATM
  • Consumer - nanny cam, door cam, gate cam
  • Each segment has it’s own market share leaders and followers
Each segment has it's own market share leaders and followers
Video Surveillance and the Distribution Channel
Video Surveillance systems companies tend to be
  • High end systems providers.
  • Regional systems integrators.
  • Technology ingredient suppliers.
  • Subsystem suppliers.
Hard Disk Drives for Video Surveillance Key Purchase criteria
HDD Key purchase criteria :
  1. Reliability (field service calls are really expensive, and lost data is a big problem).
  2. Capacity ( there are limited HDD slots and competition forces max storage space per slot).
  3. Compatibility – especially need to avoid raid controller time-out during extended HDD error correction. Dropping a second of video is OK, having a drive dropped from the raid controller is not acceptable (Samsung Command Completion Time limit is a critical feature here).
  4. Vibration tolerance.
  5. Low power, quiet and cool.
Video Surveillance Infrastructure Elements - Analog (PAL or NTSC) Co-Axial cabling based
Video Surveillance Infrastructure Elements - Analog (PAL or NTSC) Co-Axial cabling based
Typical analog surveillance system the key component is the server video adapter. Works well with existing analog installations, adds the capability of multiple display users (including web-based clients). Disadvantages include limited possibilities of camera control (remote pan, tilt, zoom is possible but expensive), and requirement of dedicated analog (coaxial) cables from each and every camera to the video adapter. Inexpensive consumer grade systems are nearly identical to the diagram above with the exceptions that they frequently employ inexpensive indoor/outdoor IP based cameras, and use a PC based platform with internal disk drives rather than a server with internal disk drives.
Video Surveillance Infrastructure Elements - IP based
Video Surveillance Infrastructure Elements – IP based
Typical digital video surveillance system. The backbone is an Ethernet digital network. Works well for new installations. Supports multiple display users (including web-based clients). Supports camera control (remote pan, tilt, zoom control is delivered over ethernet). Key advantage is wiring flexibility, optional wireless IP, Power over Ethernet (POE) removes the requirements to install AC power to every camera. Facilities upgrades are considerably easier and less expensive. For small configurations, network design and set up is straightforward. For larger configurations, network configuration does require some technical skill. Embedded environments such as ATMs are nearly identical to the diagram above, except the IP cameras transmit compressed video data over wide area networks using TCP-IP.

  • Cameras
    From the IP capable Camera -
    Typical (middle grade resolution) camera board will accept Composite Video in.
    Translate to digital: 704x480 (RGB) 30 frames per second
    This translates to~5.4 Mb/s.
    On-board digital compression compresses down to~2.4Mb/s converts to JPEG or MPG-4
    On-board Ethernet (usually POE) then transmits video content over Ethernet to a server
    Cameras-increasingly IP-based, increasingly connected simply with POE (Power over Ethernet)
    High end cameras include pan and zoom control capability
    High end cameras can be high definition using the 1080 industry standard
  • IP networks
    Most cameras are no longer analog (Co-ax).
    Most cameras connect via Ethernet.
    Either wired or wireless, often using Power over Ethernet (POE).
    They use typical (often existing) Ethernet infrastructure 100-base-T over cat-5.
    Surveillance systems use typical Ethernet routers and switches, for high end networks, these are separate from IT data networks.
  • Servers
    Servers simply connect to the network switch and handle storing data, displaying data (both real-time and previously recorded data). In low end surveillance systems, servers also handle graphics display and control. In high end surveillance systems, servers just manage data and stream MPG-4 or JPEG data to other workstations which handle display and control.
  • Monitors
    Normal NTSC capable monitors, often banks of multiple monitors are used. Increasingly HD monitors are employed often with multiple camera streams simultaneously displays. Users sitting in front of monitors generate I/O requests (rewind, play, zoom etc) so every monitor should be calculated as an I/O load on the storage subsystem. With the advent of web-based systems there can be many, many simultaneous users, creating a random workload.
  • Storage
    Storage requirements vary widely
    High end - casino, government, campus, banking, airport-large arrays, multiple 3U
    Mid-range - office, health care facility, warehouse, condo, hotel-large arrays, 3U
    Low end-small business, residential-small 1U or server tower based disks
    Embedded - ATM - usually individual cameras send surveillance video over IP to a server and storage array - surveillance from many arrays is aggregated to single storage arrays (multiple 3U racks)
    Consumer - nanny cam, door cam, gate cam – usually written to a home PC or written central storage over internet
    Typical storage - http://www.rorke.com/av/surveillance-storage/index.cfm
    Typical 3U (Rorke Data)
    Typical 3U (Rorke Data)
  • WorkLoad on the Disk Drive
    Each camera typically writes to a separate file. A camera creates a data stream of 128-256 blocks (512b), this varies depending on levels of compression. Usually transmitted over TCP (creating data stream will start and stop due to optical activity on the camera and due to compression. These are written sequentially to a file. On low-end surveillance systems, drive A is filled, then drive B is filled, then drive C will fill and then back to drive A. Typically 30 days to 60 days of recent video is retained. Higher end Video Surveillance systems store data on a RAID 5 volume. Read and Write Caching is typically ON. Buffering is typically enough to store 2-4 seconds. Drive errors can cause dropped frames, and dropped frames are acceptable in all but the highest end Casino and Military surveillance systems. the work load on the drive is non-continuous and random. Each camera writes to a separate file, in addition to cameras, human viewing creates more random requests, and sequential updates of registry block and updates of data table creates even more random requests.
  • Adjacent Applications
    Increasingly adjacent applications play an important role in video surveillance
    Example - at cash register, a surveillance camera is positioned over the cash register. There surveillance footage is time indexed. The Point of Sale transactions in the database are also time indexed. There is a transaction report that looks for anomalies (cash transactions of more than$100, checks cashed, big transactions late at night) The anomalies are then matched to the video footage for management review. This is one example of dozens of creative uses of digital video for security, for access control, for traffic, for manufacturing production, for shipping and receiving and for homeland security.
    • Badge scanning, card scanning
    • Door lock (badge) access control
    • Signature capture, comparison
    • Biometric/fingerprint, voice matching
    • Facial recognition
    • Point of Sale
    • Credit card processing/ATM/Banking
    • Shipping and receiving software systems
    • Traffic monitoring, traffic violation cameras
    • License plate recognition, parking deck
    • Cargo inspection, parcel inspection, sometimes in conjunction with RFID
    • Manufacturing systems to count and measure tolerances
    • Web-based video distribution
    • Camera control
    • Alarm system (including AoIP)
    • Public area surveillance combined digital signage
    This is important because adjacent applications serve to increase the value of the surveillance footage, and will therefore drive the need for more storage. It’s important to recognize these adjacent software situations while selling, and go for the ones with high value add.

SATA, the relevance of Queuing and Forced Unit Access
The SATA II protocol includes First Party Direct Memory Access (FPDMA) queuing which includes a thing called Forced Unit Access (FUA). Each queued data transfer includes a header, the header includes a Forced Unit Access (FUA) bit. When the FUA bit is set, the disk drive will commit the data to the media before returning success status for the command. If the FUA bit is un-set, the disk drive will write to cache and confirm write complete. Queuing in general will help these types of "value added" surveillance applications. Data can be queued first, video can be queued at a lower priority. Dropping a few video packets here and there is not harmful, dropping data packets is not acceptable. With SATA II FPDMA Queuing and FUA, an application can selectively apply FUA on a packet by packet basis to force data such as a financial transaction to write to disk and can allow the video to simply write to cache. This has the effect of allowing read and write caching to be turned “ON”, taking advantage of the F1R 32MB of cache to speed up performance, and at the same time providing security of forcing data to be written all the way to media. This technique has value if the video and data are on the same drive, or on separate drives.
Conclusion
Video surveillance has demands that are meet and often exceeded by Samsung enterprise disk drive business. The Samsung F1R RAID Class drive includes
  • high reliability - 1.2 Million hours MTBF in 24 x 7 environments
  • high capacity - 1TB, 750GB, 500GB, 320GB, 250 GB
  • 32 MB cache for high performance and fewer dropped frames
  • built for demanding Video Surveillance environments
    - vibration sensor for excellent performance, fewer read/write retries
    - SATA II industry standard with FUA, and 3Gb/s
  • Low power - 7.8 W, best in class power
  • Low heat emissions, heat is the enemy of drive reliability and worth attention
The video surveillance market is expanding in the area of basic surveillance for institutions, businesses, secure storage, malls, restaurants, etc and video surveillance is expanding with computer applications integrated with surveillance cameras; examples include point of sale, manufacturing, shopping traffic, street and motorway traffic. These increasingly sophisticated uses of surveillance related technologies drive the need is for high capacity drives (500G, 750G, 1TB). Key customer requirements are met and often exceeded: high capacity, reliability, low power, cool, quiet, vibe tolerance. Samsung F1R SATAis an excellent fit in video surveillance environments.
SATA, the relevance of Queuing and Forced Unit Access
  • http://www.synnex.com/security/index.shtml
  • http://www.synnex.com/security/hp_dvr_intro.shtml
  • Processors – MPEG4/JPEG/ JPEG2000
  • ACTi NVR 64-channel recorder software app
  • Aigis Mechtronics CCTV capture
  • American Fibretek – video port switching (POE)
  • American Science and Engineering Omnivie cargo inspection
  • Arteco vision systems
  • Bosch DVRs for security systems
  • Dahua – dahuatech.com Hanzhou China
  • Digi-IT Inc.
  • Pelco
  • Dallmeier
  • Samsung (cameras, monitors, video door phone, displays, disk drives)
  • HIK Vision
  • Dedicated Micros DVRs
  • Digimerge technologies DVR
  • Electronics Line USA ETV DVR
  • Inrevium high performance JPEG LSI
  • Matrox imaging - simultaneous capture of 16 independent audio/video sources
  • Texas Instrument DSP
  • Mitsubishi 16 channel DVR
  • Questions?? Please write h.smith@sisa.samsung.com