New communications technologies are appearing so fast these days, it's often difficult to sort out the actual applications and benefits from the hype. Voice over Internet Protocol, or VoIP, is one of those technologies and the Lab's Networking and Telecommunications Department is frequently asked what we are doing in this area. To keep LBNL staff informed of our progress and the issues at hand, here is an overview of VoIP and the issues involved in deploying it in an institution such as the Laboratory. Due to the complexity of this topic, only the most significant issues are discussed.

What it VoIP?
VoIP is one of several terms used to describe Internet Telephony and refers to a category of hardware and software that enables people to use the Internet as the transmission medium for telephone calls. For users who have free, or fixed-price Internet access, Internet telephony software essentially provides free telephone calls anywhere in the world. To date, however, Internet telephony does not offer the same quality of telephone service as direct telephone connections. Internet telephony products are also sometimes called IP telephony or Voice over Internet.

Background
VoIP was initially deployed to provide lower communication costs by eliminating or substantially reducing long-distance charges. By utilizing VoIP applications, voice communications could be transported over the data network, eliminating the need for separate voice communication trunk lines.

The implementation of VoIP services has been successful for small businesses, as well as large corporations with small remote offices (and resulting high toll charges for telephone connections). However, domestically VoIP has not made significant inroads within large centralized corporations or government agencies with in-house communication switches. This is primarily due to the high implementation cost (a full-scale VoIP installation at LBNL would be about $4 million) and the fact that many corporations and government agencies have negotiated long-term contracts with long distance carriers at favorable rates. To date, the only single site domestic installation is at Cisco Systems' facility.

In the foreign marketplace, VoIP is gaining more acceptance, such as a system supporting over 15,000 users at a government agency in New Zealand. In foreign markets with substantially higher telephone costs, companies are better able to justify the large up-front investment.

VoIP Impetus and the Implementation
Having a bit of background on telephony is an important aspect of understanding the VoIP application. Phone systems are connected by large switches that link two telephones when a number is dialed, as if to run wire point-to-point. Imagine the phone switch as the center of a many-pointed star of phones, with each point-to-point connection running through the central switch.

Touted as a replacement for regular phone switch systems, VoIP uses the existing network infrastructure of wiring and Ethernet switching hardware. This eliminates the need for central phone switches. Proponents cite other advantages such as an end to expensive phone-switch maintenance efforts, eliminating specialized telephone staff, and easier connections to off-site buildings.

On the surface, these advantages sound appealing. However, they oversimplify a complex set of issues. In fact, installing a VoIP system in a facility such as LBNL encounters a number of issues similar to those faced when centralized phone systems were first installed. Here are some of those issues:

Additional network infrastructure must be built. Accommodating VoIP requires scaling up the network to support additional ports, which requires additional network hardware. Existing hardware needs to be upgraded to support in-line power to power the telephones. All telecommunications closets must be equipped with UPS. In addition, redundant hardware must be installed in all the closets. Instead of having one centralized redundant switch, there will be hundreds of redundant pieces of hardware located throughout the site.

The network must support Quality of Service (QOS). For reliability, the Lab will have to ensure that the network backbone supports Quality of Service (QOS), which assigns priority to certain types of network traffic, and that there is sufficient bandwidth to support phone calls throughout the enterprise-at a minimum. QOS is not yet available, so Lab employees could not make VoIP calls on the current network infrastructure without the risk of interrupted calls. Additionally, QOS implementation further complicates network infrastructure, and the replacement of enterprise network routing equipment may be required.

The new VoIP phones will require power. Phones require power that today they obtain from the phone switch line cards. VOIP phones must be provided by local building power or through new Ethernet switching hardware equipment with power.

VoIP phones must be within 100 meters of the nearest Ethernet switch. The distance limitation of 100 meters effectively decentralizes phone-switching equipment by distributing it throughout the site. This creates more points of failure and increased mean time to repair. This could require additional phone/network closets.

A VoIP phone system processor must be installed and managed. Just as is the case with a phone system using centralized switches, there must be a centralized VoIP processing system to manage calls, record call details for administrative purposes, and to provide VoIP voice mail.

A VoIP interface to the local telephone company must be installed and managed. The Lab would require a staff who understands telephony and its technical nuances to provide a VoIP interface to local and long distance carriers. This is required to establish call routing rules, to track and negotiate lowest price contracts, for in-house billing, and a plethora of other things very much the same as we do now.

VoIP currently has very limited system management tools. Call processing, real-time monitoring, event alarms and logs are in their infancy.

The phone system is more reliable than the network. The current phone system's reliability is 99.999 per cent, which equals about five minutes of downtime per year. The industry norm for Local Area Networks is 99.98 per cent, which translates to about eighty-eight hours of downtime per year. (Numbers are based on statements from Cisco Systems and Northern Telecom.)

The Bottom Line
So, should we deploy VoIP at LBNL? The Networking and Telecommunications Department believes that the Lab should not at this time. Remember, a phone system is expected to be reliable, cost-effective, low-maintenance, and to serve as an adjunct to public safety. VoIP provides no extraordinary capabilities and addresses none of these issues. Finally, there are a plethora of other issues that stand as roadblocks to enterprise-wide VoIP implementations and no solutions appear on the horizon.