September 1999

A THREE-PIECE FOUNDATION FOR NEXT-GEN VOICE NETWORKS

BY BARRY CASTLE

Voice has always been mission critical, but for the increasing numbers of businesses deploying CTI solutions in conjunction with e-commerce, the converged network will actually be the business. For these organizations, network downtime will have a direct link to lost earnings, and building a network on a poor foundation could make the business vulnerable.

Certain building blocks need to be put in place in order to make the transition to a converged voice and data network. Some of these can be established with simple device upgrades, but others represent fundamental shifts away from current practice and will require detailed analysis by each individual company. This article provides an overview of three of the most critical foundation stones that companies need to put in place now to ready their data networks for voice. All companies need to take note of these issues as they start to build the network foundation for next-generation voice networks.

QUALITY OF SERVICE
One of the fundamental challenges of converged networks is ensuring a predictable quality of service. How can voice and data share network resources like switch infrastructures and line capacity in such a way as to guarantee that time sensitive traffic is delivered on time? If e-mail, Web browsing, and database queries all share the same network as telephone conversations, how can an appropriate voice quality be maintained?

Throwing bandwidth at the problem is not the answer: contention will still occur. In practice, a queuing system is required so that real-time traffic is handled before non-time-sensitive traffic. IEEE 802.1 p and q are mature industry standards that enable just such a queuing and prioritization system to operate by supporting multiple classes of traffic on the network.

Many vendors in the industry will claim that QoS only needs to be established in the core of the network, in the “control center,” or at best at the closest switch to the edge of the network. 3Com has a very different view. It is our conviction that a centrally based QoS policy will not provide a true view of the end user experience. While the network itself may look good from the center, an accurate picture of the QoS requirements of users will only be gained by extending QoS monitoring and intelligence to the desktop. Central policies will be more effective once the full picture is provided in this way.

Many vendors in the industry are introducing IEEE 802.1 p and q into their products, so companies may find that some of the necessary QoS-ready components are already installed in their networks today. The result — application-aware QoS all the way to the desktop.

Beyond local area networks (LANs), in the wide area network (WAN) arena, standards for ensuring QoS have yet to be widely adopted by the industry. More competition, coupled with bandwidth excesses and the widespread use of IP as the common denominator for traffic, will eventually lead to the creation of a spot market for voice. When making WAN voice calls or data transfers, you may be dealing with multiple carriers without knowing it. This clearly leads to concerns about how to reserve bandwidth for real-time voice calls and ensure quality of service. As we move to mixed carrier environments, we need to have mechanisms in place to ensure QoS from LANs over WANs to other LANs.

Network managers establishing corporate policies for network voice quality across the LAN and WAN will need information about the applications using the network. Without the kind of application-aware network infrastructure described here, dynamic policy management for next-gen voice will be almost impossible.

TOTAL APPLICATIONS AVAILABILITY
The issue of reliability is also critical to convergence. Recognizing that data networks have become increasingly mission critical to businesses, vendors have been building network architectures that eliminate all possible points of failure. In fact, if it is to carry enterprise voice communications, the data network must deliver five nines availability, that is, a network that is available 99.999 percent of the time. If your directory server goes down, you wouldn’t want to find that your business could no longer make phone calls, for example. This high level of availability requires careful planning and a comprehensive, end-to-end strategy throughout the network.

High availability doesn’t just mean preventing failures. For business-critical applications, adequate response time and protection against performance degradation can be as important as system uptime. In converged voice and data networks, performance degradation may be as devastating and costly as hardware failure, since degraded network performance can result in unacceptable transmission quality, loss of information, and dropped connections. Moreover, in the quickly growing area of e-commerce applications, response time can make or break business transactions.

High-availability networks must have three essential characteristics:

Resiliency — The ability of network and other devices to apply preventive safeguards as a way of heading off degradation or failure.

Redundancy — The presence of multiple units of critical hardware devices. Critical resources such as enterprise switches should redirect traffic around failed devices. Redundancy can also be applied to switch fabrics, power supplies, interface modules, servers, and other components.

Manageability — Network management can help identify critical resources, traffic patterns, and performance levels. It can be used to configure device-error thresholds, set corporate policies, and deliver sophisticated reports that show end-to-end results.

High availability must clearly be an end-to-end network goal. Companies can start work now and put measures in place to help ensure 99.999 percent availability as the minimum foundation for converged networks. Products become smarter, more “bullet-proof,” and more manageable every day, but it is still up to each organization — and each network manager — to commit the human and capital resources needed to achieve total applications availability as they transition to next-gen voice.

SECURITY — EVERY PRECAUTION AVAILABLE
Security during communication is another fundamental requirement for successful deployment of convergence. Although traditional telephony communications can be intercepted, data networks have suffered from some very high profile breaches of security. As a result, the current perception is that data networks are less secure than voice networks. Thus, even though the current level of fraudulent use of voice services would indicate that data networks are no less vulnerable than traditional systems, the transition to converged communications does provide an opportunity to improve security.

Encryption provides the necessary support for secure voice, video, and data communications, but historically, encryption introduces delays if deployed on today’s computer systems. Anticipating an increased requirement for real-time encryption/decryption associated with e-commerce, 3Com and Microsoft have co-developed a solution for real-time processing of encrypted data. (Readers interested in more information on this solution should visit 3Com product guide)

Another enabling technology for secure communications is Virtual Private Networks (VPNs) which effectively turn global IP networks into secure channels of communications. This is ideal for dispersed teams like teleworkers or multiple branch offices that want to communicate across the Internet without loss of security.

Finally, the firewall plays a crucial role in protecting systems and information from external attack. As already mentioned, many end users are concerned that the data world is more vulnerable to attack than voice systems. Firewall vendors are aware of this concern and are building combined firewall/gatekeepers to support combined voice/data systems. Here again, the concern is to deliver full functionality without requiring that customers take a step back in terms of features — clearly secure voice communications is a mandatory feature.

Businesses concerned about ensuring the reliability of their data networks should take every precaution available, and thereby strengthen the basis for future moves into voice.

CONCLUSION
Once you have established firm foundations in the three critical areas of edge-centric QoS, total applications availability, and security, you will need to bring all of these things together in one total policy management procedure for your entire organization.

Much debate exists in the industry about how to establish policy management for the new converged network. Yet there will be little point in establishing far-reaching policies if the network itself is not built with the necessary QoS support, and by the same token, network managers will need the assurance of total availability and security as their organizations transition to next generation voice communications.

Barry Castle is director of voice solutions for 3Com Corporation and was previously vice-chairman of ETSI Tiphon, Europe’s IP telephony standards committee. As part of this role, he chaired a working group on Quality of Service for IP telephony. For more information on 3Com Corporation, visit the company’s Web site at www.3com.com.