Navigating Lans, Mans, And WansHow prepared are you to deliver projects across IP and ISDN networks? Connecting multiple endpoints, troubleshooting the installation, and providing on-site logistics are all a part of balancing qual 2/01/2005 5:41 AM Eastern
Navigating Lans, Mans, And Wans
How prepared are you to deliver projects across IP and ISDN networks? Connecting multiple endpoints, troubleshooting the installation, and providing on-site logistics are all a part of balancing quality of service and security.
If videoconferencing is supposed to be the next best thing to being there, then duplex mismatch is a jarring reminder that even virtual baggage can be mishandled. “You'll hear the person speak and then see the mouth move,” says John Pfleiderer, video infrastructure and services coordinator at Cornell University in Ithaca, NY. “It's sort of the look of a poorly dubbed foreign film.”
Ferreting out the culprit requires new skills and know-how for today's broadband networking technologies, where “IP” and “Gig-E” are fundamentals like “pixel” and “SPL” are in AV. The new terms creeping into AV reflect the industry's evolution. “The process that began with the control of AV equipment over the network is now becoming more directed toward moving the AV content itself,” Pfleiderer says. “It's becoming more important that AV personnel understand the language and network topologies.”
Although there are dozens of network configurations, they can be broken down into three types: local area networks (LANs), metro area networks (MANs), and wide area networks (WANs). For more details about these different types of networks, see the “Standards 101” sidebar as well as the diagram on page 32.
An AV application might use one or all three types, depending on where the participants are located. For example, a videoconference in an office campus might use only the LAN, but if it involves buildings across town or across the country, the multimedia traffic will run across MANs and WANs, too.
All three use wired and wireless technologies, but the most common varieties are wired — fiber optics and copper — because they have the bandwidth and reliability necessary to support enterprise-class AV services. Wireless has its niches, however. One example is 802.11 — also known as WiFi — a LAN technology with a range of a couple hundred feet. Its appeal is flexibility and lower cost. “It costs a small fortune to run cables in a hotel,” says John Lopinto, president and CEO of Hauppauge, NY-based computer video technology and fiber optic transmission product manufacturer Communications Specialties. “The unions come in, and that's the end of it.”
That's why Lopinto's brother-in-law, who's in charge of AV at the New York-based investment firm Lehman Bros., is exploring WiFi as a faster, cheaper way to set up a hotel for events such as analyst conferences. The catch is that although new versions of WiFi — such as 802.11a — support data rates of up to 54 Mb/s, half of that bandwidth is eaten up by “Get-it?-Got-it” type acknowledgements between the transmitter and receiver that check whether data were lost en route. The remaining bandwidth typically drops even further — or in some cases altogether — based on factors such as interference and the distance between, say, the projector and the display. “Although these wireless technologies have their use, they're not the panacea that everyone tends to believe that they are,” Lopinto says.
On the wired side, the biggest trend is the shift away from Integrated Services Digital Network (ISDN) as the medium of choice for videoconferencing traffic in North America and, to a lesser extent, in Europe. By some estimates, networks based on Internet Protocol (IP) technology will begin to carry more video traffic than ISDN by mid-2005. “This applies broadly throughout larger corporations with mixed applicability to smaller companies,” says Jim Smith, a systems engineer with Polycom, a Pleasanton, CA-based maker of conferencing equipment. “Polycom has seen a strong push to deploy all communication devices — desk telephones, videoconferencing, fax, conference phones — over the IP networks. This seems especially prevalent in the financial community.”
The main reason is cost. IP makes more efficient use of each network link, so service providers can pass on those savings in the form of lower rates. In the case of private IP networks — such as a company-owned WAN linking multiple offices in a region — multimedia can piggyback on the existing infrastructure, reducing overhead costs.Working with IT
As the cost of transporting audio and video drops, one thing goes up: the chance that AV professionals will have to deal with more IT-related issues, such as getting videoconferencing traffic past corporate firewalls. Those issues are reflected in recent vendor moves such as Tandberg's May 2004 acquisition of Ridgeway Systems & Software, whose firewall-traversal technology is beginning to show up in some of Tandberg's IP-video products.
But even as AV equipment adds features that make it easier to work with LANs, MANs, and WANs, it helps to have at least a basic understanding of how the networks operate. That's useful not only for troubleshooting but also for working with IT departments, which oversee the networks that the company owns and leases. After all, it's tough to communicate with people from another culture without learning their language.
“It's becoming more important that AV personnel understand the language and network topologies used in TCP/IP networking,” says Pfleiderer, CTS-D, whose background includes broadcast production and classroom technology design. “In my experience, it's been useful to do the background study needed to better understand this environment.”
One example: An enterprise can configure its routers to give video traffic the highest priority, but you can't ask for that if you don't know that routers even exist, what they're for, or who controls their settings.
Doing your homework also helps avoid conflicts such as turf wars, which can occur the deeper that AV pushes into IT. “I don't see as much head-butting as there was a few years back,” Pfleiderer says. “What I see now is a trend where IT professionals are more willing to understand our needs as we hang network-controlled devices and gradually have more content delivery devices on the IP network. I believe that if we understand their concerns, which are rooted in bandwidth consumption, security and quality of service (QoS), this converged environment will steadily improve.”
A basic understanding of networks also helps when dealing with third parties, such as a service provider that provides the MAN or WAN. One thing to look for is service-level agreements (SLAs), which cover network performance. However, it's important to understand that SLAs alone don't guarantee a flawless AV experience, especially when multiple networks are involved. “Sprint offers SLAs on both voice and data transport,” says Martha DeGraw, manager of product planning and integration at the Overland Park, KS-based company's Business Solutions division. “However, Sprint doesn't guarantee quality of the individual endpoint experience due to the variability of the endpoint technologies and transport connections. Sprint's video-bridging services are designed to accommodate multiple networks and endpoint types and is the primary reason companies require video-bridging support from a carrier rather than trying to do it themselves.”Real time is no time for errors
The LAN typically is the easiest environment to control because it's limited in range and owned by the enterprise. Larger companies sometimes own a MAN or WAN, but they're typically leased from service providers such as Verizon Communications.
Controlling QoS in MANs and WANs is trickier because often there are more variables, such as different network technologies and multiple service providers. “I think that the lack of control of networks that are beyond the LAN poses the biggest problem to moving video and audio,” Pfleiderer says. “Dissimilar networks that are running at less-than-friendly speeds for video and audio delivery are still an issue.”
Standards can help. One example is multiprotocol label switching (MPLS), where a router labels incoming packets, and subsequent routers use that label to determine how to forward them. MPLS is one way to ensure that delay-sensitive IP traffic — such as videoconferencing — gets the bandwidth and priority it needs.
A network's security features also can affect performance. “The biggest and by far the most insidious characteristic of the networks that affect videoconferencing is the widespread deployment of firewalls throughout corporate IP networks,” Smith says. That problem highlights the need for working closely with IT departments to determine how to balance QoS and security.
Other common factors that affect QoS include:
- Packet loss – Videoconferencing is most sensitive to dropped packets, and it's here that good quality endpoints can make the difference by overcoming at least some network deficiencies. “There is no opportunity in real-time communications to request a re-transmit of lost information, so the endpoints have to do the best they can with the available quantity and quality of information,” Smith says.
- Jitter – Signals can get out of sync due to factors such as the distance between two endpoints, creating jitter. Again, the quality of the endpoints can make a difference. For example, some units now include ample data buffers in order to mitigate jitter's effects.
- Delayed packets – Although latency is less of an issue, it's still wise to consult the IT department and any service providers to determine whether delays could reach the point that they become noticeable to end-users. “Videoconferencing itself has between 100 and 500 ms of delay, depending on the processors used, algorithms of compression and coding efficiency,” Smith says. “So adding 20 to 50 ms of network delay is really no big deal.”
Those factors vary by network type. For example, all other things equal, ISDN typically has the lowest latency, jitter, and data loss of the networks used today. That's why it carries the lion's share of videoconferencing traffic. “ISDN networks are still very reliable and provide a more consistent experience,” DeGraw says. “But when there are network issues, the calls can drop, and it takes several minutes to bring the call back up.”
An emerging option is Ethernet, which is typically considered a LAN technology but increasingly is used in MANs and WANs, too. “Although some service providers have been offering those services for at least 10 years, the trend has really taken off over the past 12 months,” says Ralph Santitoro, co-chair of the technical marketing committee at the Metro Ethernet Forum, a trade association based in Irvine, CA.
Ethernet is attractive for its speed and flexibility. For example, the version known as “Gig-E” supports up to 1 Gb/s — enough bandwidth to keep pace with trends such as the shift from XGA to SXGA to UXGA resolution. Ethernet also is available at different levels to match the needs of a particular application, and depending on the service provider, bandwidth can be ramped up according to the application's needs and then scaled back to normal levels after the event. “They can say, ‘Today between 3 p.m. and 5 p.m., I'm going to do a videoconference, and I want to add a couple megabits,' and it's activated in 15 to 30 minutes,” Santitoro says. “That's impossible to do today with frame relay and asynchronous transfer mode (ATM).”
There are limits to self-service. For example, rather than trying to cobble together LANs, MANs, and WANs on their own, some AV and IT professionals will outsource everything that's outside their immediate control. “When Cornell needs to go off campus with a streamed event to connect a large number of endpoints, we hire out to a content delivery network (CDN) and let these companies handle the heavy lifting and managing the edge servers for a public Internet delivery project,” Pfleiderer says.
Test to impress
It goes without saying that testing is important, yet it bears repeating because even people with extensive experience in IT haven't seen it all. One key link is the connection between the enterprise and a third-party network. “This should be tested at peak times of the day so that you can accurately judge the performance for live video distribution over several audience levels of compression, if needed,” Pfleiderer says. “Typically when I'm configuring a test run, we'll establish three levels for audience delivery: one at a low bandwidth (56 kb/s), medium bandwidth (150 kb/s), and higher bandwidth (256 kb/s). The test run needs to be successful at all bandwidths for all to be considered as reliable.”
Some network service providers are certified by endpoint manufacturers to troubleshoot installations and any problems that crop up during conferences. That can be helpful because they have experience sending multimedia traffic over a variety of network and equipment types, and because they have tools that can see deep into the networks in order to track down a problem's source.
“Our most recent challenges have been in supporting conferences with H.264 applications,” says Sprint's DeGraw. “When multiple endpoints don't have the same software version and cannot view the particular file or animation within a presentation, they generally blame the service bureau. Having a test call prior is the best way to work out some of these bugs.”
But isn't H.264 supposed to be the latest and greatest? The answer is yes; however, there's a catch. Although the codec has been done for more than a year and most videoconferencing vendors are already supporting it, that doesn't address legacy systems so problems can arise.
Tim Kridel is a freelance writer and analyst who covers telecom and technology. He's based in Kansas City and can be reached at firstname.lastname@example.org.