November 1998

We asked several industry-leading vendors for their views on the Internet telephony industry. Their responses appear below.

• Gordon VanderBrug, Executive Vice President, VIP Calling Inc.
• Mark Baker, CEO USA Global Link, Inc.
• Howard Bubb, President and CEO, Dialogic Corporation
• Peter S. Buswell, Franklin Telecom President/COO
• Dr. Henrik Sorensen, VP Advanced Technology, elemedia
• Steve Jaroski, Bellcore VP and General Manager, Voice over Packet Solutions Delivery
• Prabhu Kavi, IP Business Marketing Manager, Core Systems Division, Ascend Communications, Inc.
• Tom Evslin, Chairman and CEO, ITXC Corp.
• Al Bender, VP and General Manager, VoIP Network Solutions, Nortel (Northern Telecom)
• Peter Sommerer, Chairman, Vienna Systems Corp.

Gordon VanderBrug, Executive Vice President, VIP Calling Inc.
Today, Internet telephony is implemented with small computer-based gateways that provide up to two T1s of capacity. To deliver high-capacity quality service, a provider must integrate multiple devices, including routers, CSU/DSUs, multiplexers, DSX panels, and remote access devices. In addition, specialized tools are required to rigorously monitor the network to assure good performance. We have found that toll-quality Internet telephony is possible today, provided one is willing and able to do this level of painstaking systems engineering and network monitoring.

On the other hand, Internet telephony still needs to take significant steps before it becomes "fully fledged" and assumes the dominant position in the marketplace. The industry needs:

1. Gateways with thousands - not dozens - of lines.
2. More robust gateway hardware and software approaching the reliability of tandem switches.
3. A network supporting quality of service with priority for real-time data.
4. A price per port that is reasonably close to that of circuit switching.

These issues require more urgent attention than either interoperability or the development of value-added services.

[return to the top]

Mark Baker, CEO USA Global Link, Inc.
The key to growing the Internet telephony market from hacker-domain to carrier-class is integration. This integration will be apparent in two areas. The first is internally driven -a seamless, worldwide fully integrated network. Although IP-based networks have greater potential for efficiency, cost-effectiveness, and functionality, circuit-switched networks will remain a dominant part of the telecommunications landscape for many years to come. Rather than investing in an IP-based network which would exclude other transmission modalities, the intelligent approach will be to acquire an integrated SS7/IP-based network. Our industry depends on equipment manufacturers and vendors, such as Siemens and 3Com, to develop and support new, integrated solutions. In addition, we require worldwide interconnection standards to make IP-telephony service transparent to customers.

The second area of integration is externally motivated. Technological innovations like IP telephony have given birth to sophisticated global customers who are exposed to and demand a broader range of service, all at increasingly lower prices. They require fully integrated service through a single delivery channel. Carrier-class Internet telephony service providers will profit from the expanded world market by offering their customers a unified package of telecommunications services incorporating traditional service with new, innovative services while maintaining the flexibility to adapt to newer offerings as they become available.

Carrier-class Internet telephony requires robust and interoperable technologies designed around worldwide standards. On this foundation, we can deliver an array of advanced and convergent communications tools to the rapidly expanding global customer base.

[return to the top]

Howard Bubb, President and CEO, Dialogic Corporation
Many things will fall into place to deliver full-fledged carrier-class IP telephony gateways. Certainly a set of carriers are raising the capital and putting in the infrastructure to support massive IP telephony deployment. There are active discussions within the myriad standards organizations about protocols to integrate IP telephony gateways and IP backbones within existing and evolving intelligent network architectures.

Protocols and standards are in place and have first been implemented in small-scale systems. The equipment vendors must now deliver equipment that meets the availability, density, and cost requirements of carriers. A key piece of the solution for those of us in the open systems camp is CompactPCI. CompactPCI offers both important features for high system availability, with standard components to ensure competition among providers and cost-effective solutions. In a fortunate piece of timing, CompactPCI solutions for both platforms and boards are coming to market just as the IP telephony industry is entering volume deployment. Combined with mission-critical operating systems from Microsoft, Sun, and SCO, these CompactPCI solutions provide fault resilient, scalable platforms that carriers are comfortable deploying.

Built around open systems, including the H.110 backplane TDM standard from the ECTF, the component community is delivering multispan CompactPCI gateway components that enable DS-3 density solutions in a single rack. The most advanced of these products include integrated PSTN connections, a full range of voice coders than can be negotiated on the fly, T.38 real-time fax, and integrated IP connection on a single card, enabling fault resilient, scalable IP gateways.

[return to the top]

Peter S. Buswell, Franklin Telecom President/COO
There are three characteristics that typically drive the adoption rate of a new technology: economics, the user interface, and user applications. Based on the value Wall Street places on a Quest, for example, when compared to any CLEC stock offering, we can witness the economics of this technology in real-time. As a straightforward ROI analysis, IP networks are a no-brainer. This economic advantage will, in and of itself, drive network creation. Before true market demand will drive the development of carrier networks, we are going to have to create IP applications that offer something more than "cheap" phone calls. We need rich, multimedia applications like Web-enabled call centers and online commerce in which the technology delivers true "caller ID" and functionality specific to IP telephony. Economics, acceptable user interfaces, and IP-enabled applications can cause the level of demand and excitement that will accelerate network deployment. Let's get creative.

[return to the top]

Dr. Henrik Sorensen, VP Advanced Technology, elemedia
For large, established carriers to truly adopt IP telephony and move beyond the limited trials that are currently being conducted, there are serious hurdles that must be navigated. An obvious one is reliability. The IP industry is still plagued by the image of an apartment on fire and a guy frantically rebooting his computer to dial 911. As the industry is moving away from PC-based solutions towards more dedicated software and hardware solutions, these problems are being solved. Other concerns that have started being addressed are security and interoperability -- and although the solutions are not ready yet, work in standards organizations is pointing in the right direction.

One of the biggest challenges facing the industry is figuring out how to successfully integrate IP telephony with existing network management systems to allow established carriers to gradually integrate these new technologies. Network management is also needed to provide the "network awareness" that is required to increase reliability and availability. The redundancy within each hardware and software component can only resolve a very limited set of problems without this higher-level awareness. Even security and interoperability are heavily dependent on network management functions to provide provisioning and protocol translation.

[return to the top]

Steve Jaroski, Bellcore VP and General Manager, Voice over Packet Solutions Delivery
Today, there are at least three variations of "Internet Telephony." Telephony over Packet (TOP) means that companies can place, complete, and receive telephone calls using a packet transport network and a high-speed protocol like frame relay, ATM, or IP. Internet Telephony refers to telephone calls made over the public Internet where the voice is digitized and transmitted using the Internet Protocol (IP). Finally, there is IP Telephony that refers to the ability to complete telephone calls where the voice is digitized and transmitted to the receiving end using the Internet Protocol.

Regardless of the variation, to move from small-scale solutions to full-fledged carrier-class, Internet telephony will need to adopt several important characteristics. First, there must be interoperability between IP networks and the PSTN. And, PSTN voice quality must be preserved. A rich set of telephony features like E911, 800 services, and call waiting services should be integrated with the carrier's ability to provide directory services, accounting, and settlements and security features. Also, the overall service delivery architecture, including the hardware, software, and data network infrastructure, must have carrier-grade reliability exceeding 99 percent, in addition to scalability and network management capability. Finally, establishing a carrier-grade OSS will complete the evolution. When we have these capabilities in place, we'll have a genuine carrier-class Internet telephony solution anyone can use.

[return to the top]

Prabhu Kavi, IP Business Marketing Manager, Core Systems Division, Ascend Communications, Inc.
Put simply, achieving carrier-class Internet telephony requires delivering service quality equal to that of the PSTN. While some niche markets can take advantage of lower service quality, such as low cost voice over the Internet, VoIP will not achieve its potential while consumers still equate VoIP with substandard quality. There are four key requirements that VoIP gateways need in order to meet the PSTN standard.

Highly available: The equipment must approach downtimes of but a few minutes per year (including upgrades).

Guaranteed quality: Every call must have the same consistent voice quality, taking into account available backbone bandwidth. If placing a new call would adversely affect calls in progress, it should be rejected.

Scalable: VoIP gateways must be able to scale from hundreds of ports up to the size of a traditional voice switch, or 100,000 ports.

Network Management System: A robust network management system is required to provision, manage, and generate billing information for a VoIP network approaching a million ports.

No equipment meets all of these requirements today. Therefore, carriers must pressure equipment vendors to meet these requirements if carrier-class Internet telephony is to become a reality.

[return to the top]

Tom Evslin, Chairman and CEO, ITXC Corp.
Internet telephony is already moving into the mainstream with scarcely a ripple. Resellers (not ISPs) are the early adopters. They are buying worldwide IP call termination from those who can provide them a footprint with consistent quality, and putting this termination in their routing tables as they do other termination options.

However, ITXC - and its competitors who are growing that footprint - are hampered by three things:

• Lack of widespread interoperability between gateways from different vendors;
• Lack of SS7 support in gateways;
• The small scale of gateways available today.

For the next ten years, IP and PSTN networks must interoperate. In order for IP users to enjoy the full features of today's PSTN (along with the many benefits of IP), gateways need to talk to switches in order to support intelligent billing, roaming, and other functions. Switches already communicate control information with SS7. Gateways must communicate with switches in the same way.

The gateways being used today are toys (suited, however, for early adoption). 96 lines is "big" for a gateway. Switches are two orders of magnitude larger and one order of magnitude cheaper per line. Most of this gap will probably be closed in 1999.

[return to the top]

Al Bender, VP and General Manager, VoIP Network Solutions, Nortel (Northern Telecom)
The creation of full-fledged, carrier-class IP telephony is no small feat. There are a number of critical areas that need to be addressed, including:

Services - Using IP telephony to avoid access charges is a business with a limited window of opportunity. Ultimately, it is the availability of new services that will drive IP telephony.

Protocols - H.323 was never designed for large, carrier-grade IP telephony. Consequently, it is being augmented with newer protocols, such as IPDC.

Reliability - Users will not accept a loss in reliability as they move from the circuit-based PSTN to IP telephony. Network architectures, equipment, and operational processes must ensure equivalent reliability.

Scalability - Service providers need to install networks appropriate for the their current business level but be able to easily scale the networks as business grows. Networks must scale up to millions of users.

Transparency - Users must be able to use IP telephony without changing their behavior.

[return to the top]

Peter Sommerer, Chairman, Vienna Systems Corp.
Traditional telephony service has numerous strengths: It is a completely ubiquitous communications technology with extremely high reliability, and has a powerful combination of sophisticated services (e.g., caller ID, 1-800, conferencing) with very low-cost and low-complexity end user devices (phones). In order to supplant traditional telephony, IP telephony must approach traditional telephony in terms of these advantages. Ubiquity and reliability of IP telephony is obviously heavily dependent on the underlying Internet infrastructure. Ubiquitous features and services, however, require an intelligent architecture that extracts call control and device control from the network fabric, such as the switching devices (gateways) and end-user devices (IP phones). Who wants to upgrade their gateway or phone software (like we do our PCs today) every time a new feature is introduced? Or worse, throw out your phone for a new one that needs more computing power?

Distributing services such as PBX services, call center services, or IP Centrex services on servers in the network will allow user service and routing information to be configured and monitored at a single point, speeding network and service deployment. At the same time, this architecture allows for lower-cost network devices and lower-complexity end-user devices. Finally, separating call and device control from the switching fabric allows easier adherence to new signaling standards as they mature, enabling easy and efficient access and interoperability to all vendors' gateways.

[return to the top]