It's just a matter of time before voice runs as just another application on data networks, but that doesn't mean the time will be right for you in 2002. VoIP products are maturing every day, but they aren't yet as mature as vendors would like you to believe. As yet, you won't have much to gain by ripping out your existing PBX system and replacing it with this new technology. If you are putting in a new phone system, however, it wouldn't hurt to consider VoIP.
Before you do anything, make sure your data network infrastructure is reliable and can provide the QoS needed for VoIP. At minimum, ensure that voice packets get first choice, and everything else comes second.
Running voice over your service provider's network may prove even more challenging. If you decide to run VoIP over the Internet, make sure all the connections are with the same ISP. And make sure the ISP will guarantee the QoS you require and that it can stand behind the guarantee. And if you're thinking about using multiple ISPs to do this, think again. You could be setting yourself up for big-time finger-pointing at best and horrible performance at worst.
If lack of reliable ISP support isn't enough; another good reason to wait is the lack of standards. In fact, vendors of enterprise PBX replacements continue to offer proprietary phone sets, and their excuse for this has been the lack of standards. Well guess what? They are about to run out of excuses.
SIP (Session Information Protocol) continues to mature, and it offers rich functionality. In addition, more vendors are offering SIP phones and participating in interoperability bake-offs. Remember, even if a vendor gives lip service to a standard, the goal is interoperable products.
Economies of scale will make mass production of VoIP phones profitable when vendors can count on the large market that a single standard will generate. Although you probably won't see the drastic price cuts this year, you will be able to ask the VoIP vendors why they don't provide native support for multiple vendors' IP phones. Expect the enterprise VoIP PBX vendors to drag their feet as they find creative ways to sell you their expensive proprietary technology. Don't let them get away with it! Hold out as long as you can. They'll get the message.
If you want to give VoIP a try, start with a pilot. Use this opportunity to make sure the system will do everything you need it to do. Test it long enough to avoid the frequent software updates and potential instability that an untested technology can generate.
You also need to make sure that you define your terms when assessing this technology. VoIP on the WAN, to replace voice trunks, is related but poses different concerns from those of a VoIP that's used to replace a legacy PBX. You can have one without the other, or you can have both together. You need to assess them as separate issues.
ATM Technology
Although ATM has been put on the back burner as an enterprise technology, it is still simmering. In fact, ATM continues to be a strong Layer 2 technology in the service provider and carrier world. This position doesn't seem to be diminishing, as seen by the number of new services that are being offered over ATM networks, including voice, DSL, Internet access, video and MPLS.
For the past several years, the song has been about the overhead costs of ATM and why enterprise customers shouldn't use it--a tune that pretty much killed ATM in the LAN. It continued to grow on the WAN side, however. Customers haven't had any other option but to use ATM if they wanted good QoS from end to end. That will begin to change slightly in 2002 with the adoption of MPLS.
While use of Ethernet is gaining acceptance in the enterprise, the world is a long way from having Ethernet replace ATM in the core. Service providers are deploying large Gigabit and 10 Gigabit Ethernet networks as they provide better bang for the buck than ATM, but for the foreseeable future ATM will not be replaced by carriers to transport their services.
Most data traffic--from frame relay to Internet--flows over ATM connections around the globe. ATM is gaining in the marketplace simply because of the rising number of DSL modems in homes and offices. Although some products use frame relay or PPP to transmit data to and from the DSLAM (DSL access multiplexer) and modem, ATM is still top dog because Ethernet can't come close to its QoS guarantees.
Even from the DSLAM to the core, ATM is the most common transport because it rides on top of SONET, which can run long distances--thereby remaining king of the Layer 1 protocols. SONET is also bolstered because Gigabit Ethernet standards can't exceed 150 km, and 10 Gigabit is not able to transmit farther than 40 km.
Enterprise customers looking for more reliable means to transmit their IP traffic can begin looking at MPLS services offered by carriers. ATM and MPLS bring the best of both worlds to enterprise customers. MPLS offers tagged data to enable more efficient routing while transmitting over an ATM network. 2002 will be the first full year in which most carriers will have MPLS services to offer to customers.
On the hardware side, vendors continue to build bigger and faster ATM switches. Major vendors, like Alcatel, will be deploying OC-192capable ATM switches early in 2002. The development in this area is aimed at the carrier and service-provider market. This means the need for ATM networks to increase bandwidth and mileage will continue to grow in 2002 but at a slower pace. Service provider and carriers will lay more fiber and increase the bandwidth of the lines they already have wherever possible.
Sean Doherty is a technology editor based at our Syracuse University Real-World Labs®. A former project manager and IT engineer at Syracuse University, he helped develop the infrastructure behind a campuswide, centrally supported applications and storage system. Send your comments on this article to him at sdoherty@nwc.com.
RSS
