Serving up VIDEO
January 1, 2008
Terminology for technology unless codified in standards sometimes is defined by each manufacturer in the way its system is configured or by security dealers, systems integrators and information technology (IT) professionals in the way they design systems.
This is especially true when different technologies and industries are blending together, such as IT and more proprietary analog security systems, and when technology is evolving from analog to digital. So it’s not surprising that definitions of a video server vary.
“I’ve noticed a lot of people are using all these different terminologies around the industry, and nobody really knows,” observes Doug Corbett, business development for Viscount Inc., Burnaby, B.C. “The device some people call an encoder is a video server. We look at it as bringing in any input video and serving up the images to a client somewhere. The word encoder somehow got mixed in.”
Fredrik Nilsson, general manager of Axis Communications Inc., Chelmsford, Mass., sees many different definitions of video servers. “We launched our first product back in 1997, which we called a video server,” he remembers. “We still call them video servers, and that’s what other companies call video encoders.”
The term server is confusing, Nilsson thinks, because to some people it means a storage device like a computer server and to others it means a device that streams video to the Internet like a Web server. “That word specifically will fade away and be replaced by video encoder,” Nilsson says of future terminology for a video server.
David Engebretson of Slayton Solutions Ltd., Chicago, one of SDM’s contributing technology writers, thinks the term server is used because most are accessed through a Web browser.
“What is commonly called a video server is an encoder,” Engebretson admits. “The reason they get called servers is that 95 percent of the products out there function as Web servers, meaning that to access them, you use Internet Explorer.”
That means a video server can be accessed from any computer if the user has its IP address. For installations with multiple servers, Engebretson recommends use of video management software that can view multiple servers on the same screen at the same time instead of accessing each one separately from a Web browser.
“The encoder is hardware â€” it’s a little box twice the size of a pack of cigarettes,” Engebretson explains. “It’s got a circuit board, a CPU and a power supply. The basic things it has are the input for the camera, the video feed, an RJ45 female socket to plug into the network, and some kind of power.
“Some devices will power over Ethernet (PoE), but many require a separate power supply,” Engebretson reports. “All some servers will do is what we call unicast, which means they will transmit to an authorized user, and the authorized user can hook up to it and get the video.
“Some manufacturers’ products will multicast using specific IP addresses that feed the video stream into the network and the network itself distributes the video,” he notes.
After converting to digital, the resolution, frame rate, brightness, contrast and other attributes of the digital video’s image can be set on the server by the installer, including the compression format, such as M-JPEG, MPEG-4 or H.264.
SHOULD SERVERS ALSO STORE VIDEO?“From our perspective, we like to define a video server as a device that provides the ability at minimum to capture and record video and then provide it back for review, as opposed to in general what I would suggest the industry considers an encoder,” asserts Bob Beliles, senior manager, market management, physical security, Cisco Systems Inc., San Jose, Calif.
“An encoder is a device which digitizes and compresses video and leaves it at that â€” that would be your basic function,” Beliles continues. “A function above that would be a device that takes the video, digitizes and compresses it, and puts it into a network packet, which is a gateway device, which allows it to connect to a LAN or WAN and ultimately to the Internet and recording platforms.”
These would include digital video recorders (DVRs) and network video recorders (NVRs), he says. Some of these may be hybrids that can have analog or IP cameras connected to them. Servers also can connect to network-attached storage (NAS) devices or storage area networks (SANs).
Matt Barnette, vice president of sales at AMAG Technology Inc., Torrance, Calif., agrees that video servers should include storage. “A server not only transforms the signal from analog to digital but also saves the data,” he insists. He adds that such a device allows moving more of the system to the edge to reduce cabling costs.
“Our idea was to put the camera out there and put a video server out at that location and be able to connect it to the network wirelessly and have it stream the video back, and you can be recording it or viewing it anywhere you have network connectivity,” he relates. “It allows you to increase the flexibility of the design as well as the usability of the system.
“There should be a significant cost savings during installation of the system utilizing the network infrastructure already in place, so you don’t have to run cables to all these devices â€” you simply connect them to the network, and your device is now online,” Barnette points out.
He adds that some advanced IP cameras are not only performing the server function, but also storing video. “So in essence those cameras have become very intelligent devices at the edge of the network,” he notes. “They are acting in somewhat of a server capacity, where they are storing the video or making some type of intelligent decision.”
To some, DVRs are all that is needed. “The way we see encoding is a little bit different,” concedes Mark Provinsal, vice president of marketing for Dedicated Micros Inc., Chantilly, Va. “We think a DVR is also an encoder with storage built into it. We sometimes treat our DVRs as video servers.”
Corbett agrees. “We provide a video server, but Viscount’s also a DVR at the same time,” he relates. “We use nonproprietary PCs for our video server, and we can turn on the encoding function by just clicking a button.” He notes that Internet protocol (IP) cameras have the encoding function built into them.
“The video server is the exact same technology in an IP camera,” Engebretson concurs.
Continues Corbett, “From there, you move up to an analog camera that has a video server built into it. That video server sometimes is called an encoder, but it’s a single-channel device.
“The next step up from that is a standalone video server that is often called an encoder and takes from one up to four channels typically,” Corbett explains.
Some video servers are mounted in racks if existing video surveillance systems already are configured to send analog camera output to centralized locations like control rooms. A 19-inch rack can have a capacity of up to 48 ports or channels.
Such a rack is made up of 12 four-port video servers, each of which does not have a housing and is called a “blade.” Blades only can function in a rack.
Other servers are strictly software that can reside on any off-the-shelf computer and perform the functions of a video server.
THE FUTUREBecause video servers are a transitional technology, Corbett expects the market for them to increase for a while and then drop off.
“As we move more and more to IP cameras, they will come to a point where these have saturated the market, and all the analog cameras that you want to change to IP have been done,” he predicts. “The newer cameras coming on are IP cameras, which are video servers in themselves.
“It’s hard to say how much time â€” IP cameras are growing so fast in the marketplace â€” but over the next five years, there will be a reasonable growth in IP video encoders, and then after that, they’ll just drop off the market,” Corbett forecasts. “They won’t be needed anymore.”
But the number of analog surveillance cameras still in use is high, Barnette points out. “Those aren’t going to go away anytime soon, so over the course of the next five to 10 years, there is going to be a need for hybrid devices to put those cameras on a network,” he concludes.
Sidebar: What to Look for in a Video ServerPete Jankowski, director of product marketing for the physical security business unit of Cisco Systems Inc., San Jose, Calif., advises keeping the quality of the digital video picture at the same level after conversion as it is in analog.
“If you’re turning analog into digital, keep the quality and frame rate the same at least as the capturing camera,” he recommends. “If you’re looking for high resolution, 30 frames per second (fps) at the NTSC standard with 704- by 480-line resolution, which is 4 CIF, that’s probably pretty adequate for an encoder to encode at the top end.”
Fredrik Nilsson, general manager of Axis Communications Inc., Chelmsford, Mass., points out that picture quality can be affected by video servers. “You might think that the picture quality from the video encoder-based system depends on the camera,” he relates. “The video encoder actually does affect the video quality.
“So some video encoders have video quality that is much better than others, but that’s something that’s difficult to list on a data sheet,” Nilsson concedes.
He recommends that dealers ensure that four-port video servers they install be able to deliver 30 fps at 4 CIF in all four ports at the same time, which he calls a premium product.
“There are a lot of products that can’t really do that â€” maybe they deliver half that performance,” Nilsson maintains. “The most common difference between different video encoders is performance.”
Jankowski agrees that four-port video encoders frequently use one chip over the four video streams to economize, so they may not be able to handle the highest frame rates and resolution over all four video streams.
“That is why you don’t see the secondary streams on a lot of those,” he explains. “When all four ports are going at a high rate, there is no head room left. You don’t want to have a dedicated digital signal processor (DSP) chip for every video stream, but you lose something, you can’t run them at full rate, and if there is any video analytics, there’s no room left.”
He recommends running video analytics back on the computer server rather than on the video encoder.
If required by the installation, a video server should be able to provide a secondary stream of compressed video in addition to a live stream of high-quality video.
“You should be able to do a secondary video stream off it and use standard compression, such as M-JPEG or MPEG-4 or H.264, depending on what your options are on storage and resolution,” Jankowski advises.
“Many of the earlier or more basic encoders have the ability to provide one video stream at an adequate or higher level of resolution, but when you ask for a secondary stream, there are some encoders that don’t have the performance capacity to provide two streams at the same compression level and frame rate, and the video would degrade,” adds Bob Beliles, Cisco’s senior manager, market management, physical security.
Beliles recommends using a server with high-performance encoding in the hardware with a DSP or an application-specific integrated circuit (ASIC).
He emphasizes that performance is key. “Why one encoder seems to cost more than another one might be a function of how robust the hardware or the processing capabilities are in a given encoder,” Beliles notes. “That can make a big difference. It may not be so clear on the data sheet, but it’s certainly well worth understanding those capabilities and getting a benchmark.”
Jankowski adds that latency also is important, especially when dealing with pan-tilt-zoom cameras that are controlled through the video server. “Latency typically is defined in the video business from the point of capture at the camera to when it’s displayed to the user,” he explains. “Latency over 200 milliseconds is very tough to control â€” if it’s more than that, it’s kind of difficult.”
Sidebar: What to Do with that VideoIn larger installations, video systems are starting to share their storage with other systems’ data. “A lot of these systems are starting to use the same disk storage,” points out Pete Jankowski, director of product marketing for the physical security business unit of Cisco Systems Inc., San Jose, Calif. “In the past, most video systems were self-contained.
“It’s starting to happen where security may be using the same storage array for access, video and other pertinent data,” describes Jankowski. “So the systems are standard, off-the-shelf components, except when you get to the actual encoder.”
The security of the hard disk drives and backup on RAID arrays is important, notes Bob Beliles, Cisco’s senior manager, market management, physical security. “You can also have software that monitors the health of the entire storage device itself,” he points out. “If there are problems with the storage device, it can fail over between two different storage devices.”
Matt Barnette, vice president of sales at AMAG Technology Inc., Torrance, Calif., says his company’s video servers use hard drives designed for laptop computers, because they are made to handle more rugged use.
Although Barnette thinks hard drive costs can be economized in redundant, hot-swappable RAID arrays where the drives can be replaced while the system is still storing data, Beliles and Jankowski suggest using a recognized manufacturer’s disk drives.