Previous Table of Contents Next


Ethernet CIF Access. CIF allows a desktop application to place voice traffic in ATM cells that are subsequently inserted into Ethernet frames by the network adapter driver for transport over the link from the adapter to switch. At the Ethernet switch, cells are extracted from the frames and sent across the ATM backbone.

CIF’s primary advantage is that high-priority traffic, such as voice, can be given the necessary QoS from the desktop across the ATM network without having to actually install ATM end-to-end in the network. Furthermore, because voice does not utilize IP, the desktop Ethernet switch can be simpler and need not include more expensive and compute-intensive routing functionality. In this way, CIF may be a potential alternative to RSVP/IP for organizations migrating to switched Ethernet to the desktop but also interested in deploying voiceLAN.

CIF’s ability to guarantee quality of service comes at a price. CIF requires installation of special software or NIC drivers in workstations to accomplish the framing of ATM cells. In addition, transporting traffic inside of ATM cells, which are in turn encapsulated by frames, entails significant overhead, reducing the usable bandwidth on an Ethernet segment to 6M-bps to 7M-bps.

CONSOLIDATION OF THE CABLING PLANT

A consolidated cabling plant that supports both voice and data is one of the primary benefits of implementing voiceLAN. With voiceLAN, the cabling plant supporting voice communications (e.g., cabling runs, patch panels, cross connects) becomes a redundant, backup infrastructure that can be removed when the voiceLAN network stabilizes.

No matter what technology is used for voice transport (i.e., ATM or IP), voiceLAN deployment requires optical fiber in the risers of buildings for backbone connectivity. Most large organizations have already installed fiber for their LAN backbone and therefore no upgrade to the cabling plant is necessary. Exhibit 1 and Exhibit 2 depict a consolidation of cabling plants through voiceLAN technology for a typical organization.

MIGRATING THE DESKTOP

The deployment of voiceLAN also entails a migration of the desktop PC to become telephony-enabled. Exhibit 3 illustrates the voiceLAN-enabled desktop environment. This migration has two components: hardware and software.

Hardware Upgrades

In a pure voiceLAN architecture, all voice calls are received via a PC and its LAN adapter card rather than via a desktop telephone wired to a PBX or voice switch. There are two alternative human interfaces for people to interact with the PC to receive voice communications: the PC itself and the traditional desktop telephone.

By using the PC as the interface, voice traffic is processed by a PC sound card and the user employs a PC-attached microphone and headset. This solution is appropriate for users who are already using a microphone and headset to keep their hands free for typing (e.g., telemarketers, travel agents, help desk operators). Disadvantages of this set-up include the fact that voice packets are processed by the PC’s CPU, potentially hampering performance for other applications that might be running simultaneously. In addition, if the PC locks up, the user’s conversation may be interrupted.

For most users the desktop telephone is still appropriate as their voice communications interface. However, in a voiceLAN solution, this phone set must be able to connect directly to the PC so that voice traffic can be received directly from the network adapter card without having to pass through the CPU. Today this can be accomplished through a third-party plug-in card.


Exhibit 1.  Legacy Voice and Data Cabling Infrastructures

Universal Serial Bus A more elegant solution for accomplishing a direct connection is the USB interface, originally developed by Intel. Within the next year, the motherboards of most new PCs will include USB interfaces as a standard feature.


Exhibit 2.  Consolidated Cabling Infrastructure.

The USB supports 12M-bps of throughput and allows USB-compatible telephone sets to connect directly to the PC without the need for an additional plug-in card. This alternative greatly reduces the cost of deploying voiceLAN. Several vendors have released or will soon release telephones conforming to the USB standard.


Exhibit 3.  The VoiceLAN-enabled Desktop.

Firewire Bus An alternative standard called Firewire — originally developed by Apple Corp., but currently being promoted by Sony and other consumer electronics companies — is also being introduced in new products.

The Firewire bus runs at speeds of up to 400M-bps, which makes it appropriate for video traffic as well as voice. This high level of performance also may make Firewire too expensive for ubiquitous deployment, particularly if voiceLAN, not video, is the driving application. PC vendors may also be slower to develop Firewire-compatible telephones because of the Macintosh’s declining share of the business market. Therefore, deploying USB-compatible phones is currently the most prudent choice for voiceLAN migration at the desktop.

Software Upgrades

To take maximum advantage of voiceLAN technology, PC-resident applications need to communicate with the PBX and PC-attached desktop phone sets. For this, a standardized software interface is required.

Most PBXs today support several such software APIs, though many of these interfaces provide translation of commands between the PBX and mainframe hosts for use in CTI applications such as call center applications. For Windows applications, most PBXs support Microsoft’s TAPI. TAPI is available for Windows 95.

Microsoft Corp. is introducing a newer API, combining its Windows data transmission API (Winsock) with its voice communications API (TAPI). This consolidated API, known as Winsock 2, makes it even easier for developers to write integrated voice and data communications applications.

Clearly, the migration to a voiceLAN architecture is made much easier for organizations planning a substantial (Wintel) PC procurement in the next six months. These PCs are considered “voiceLAN ready” because they include both USB motherboards and systems software supporting TAPI/Winsock 2 APIs. Legacy PCs would need to have TAPI/Winsock 2 software and a third-party adapter board (for handset connectivity) installed.

MIGRATING THE PBX

Legacy Telephony

Today’s PBX and telephony systems are analogous to the host and dumb terminal model of the mainframe era. PBXs are relatively inflexible, proprietary, and expensive to maintain and upgrade in the same way mainframes are. Phone sets are still the most ubiquitous desktop instrument for telephony communications, but the PC offers the most intuitive interface to advanced features. Moving from the traditional PBX model to a server-based telephony model represents the final stage in the migration to a fully integrated voice and data network.


Previous Table of Contents Next

Copyright © CRC Press LLC