Are you ready to be unified?
Unified messaging (UM) technologies integrate voice and fax services with email systems, including Exchange Server. UM products have been available since the mid-1990s but have so far failed to penetrate or captivate the messaging marketplace. However, recent developments in communications and networking, as well as changes in UM technologies, might finally give UM a shot at the big time. If you've been considering a UM solution and wonder whether the time is right, your best bet is to make sure you understand how the technology works and more important, how it works with Exchange.
What's UM All About?
UM generally refers to messaging functionality that provides access to (at the least) email, voicemail, and fax services via (at minimum) a telephone and a PC-style connection (i.e., a Web interface or an email client). UM vendors claim myriad benefits for the end user—principally, productivity gains through better access to messages and financial gains through management improvements. In most circumstances, these improvements are of greatest benefit to mobile users (e.g., salespeople who travel frequently and need access to email, voicemail, and fax services from any place at any time via any device). So why hasn't UM been a hit with consumers? Several reasons:
As you can see, these concerns boil down primarily to buying concerns (costs, ROI, and technological staying power) and implementation concerns (performance and management concerns). Knowledge is power, as the saying goes, so let's explore these two areas of concern so that you can better evaluate whether UM is right for you.
Should you invest in a technology that can be costly, doesn't generally provide hard-and-fast ROI figures, and is still evolving? The contrarian's view is that UM solutions will soon be outdated now that "smart" mobile devices offer access to voicemail, email, and (to some extent) fax services. Still, though users might have one device to access data, all that data is retained in separate logical locations. True unification isn't possible unless diverse data is stored in one location. Interestingly, this is one way in which UM is evolving: in an overall sense of communications convergence, giving rise to the phrase unified communications.
Aside from an overall concern about whether UM is here to stay, you might worry about the speed at which UM technology has evolved over just the past few years. Will the UM solution you implement today be superseded by something better tomorrow? Well, maybe—but in all fairness, the same applies to any technological purchase. You definitely want to be sure, though, that any solution you consider buying takes changing technology into account. Products you buy now must meet not just contemporary requirements but must also have a roadmap for tracking future advances.
When you take such precautions, you might decide that the rapidity of UM advancements is actually an advantage. Most of the changes that have taken place in recent years have benefited users. Take the advent of IP telephony. Before such systems came on the scene, UM implementation was hindered by the complexity of integrating Exchange servers and existing PBX systems. The vast number of communication protocols (e.g., Common Channel Signaling 7—C7, Channel Associated Signaling—CAS, Common Channel Signaling—CCS, dual-tone multifrequency—DTMF, E1, multifrequency—MF, Microsoft Foundation Classes—MFC, R2, T1) associated with PBX communications was outside the realm of knowledge of most Windows IT professionals, who also had little knowledge of the technology necessary to integrate PBX systems and Exchange (e.g., computer telephony integration—CTI—PCI cards such as Brooktrout Technology's TR1000 and Aculab's E1/T1). Furthermore, the implementation of CTI PCI cards and their associated software often caused compatibility issues that rendered Windows host systems unstable.
Now, though, most UM vendors have updated or are updating their applications to use IP telephony interfaces instead of traditional PBX interfaces. PBX suppliers and networking vendors offer a range of network devices that connect PBX systems directly to LANs, by using either plug-in CTI-interface components on the PBX or dedicated Voice over IP (VoIP) network gateways. UM Exchange servers can now communicate with PBXs simply by using TCP/IP as a network protocol and H.323—or increasingly, SIP—as a signaling protocol. Figure 1 illustrates the simplicity of such a setup.
The introduction of SIP as a signaling protocol has also brought a number of advantages. (For more information about this protocol, see the Windows IT Pro article "Understanding the Session Initiation Protocol," January 2003, InstantDoc ID 27397.) Basically, SIP provides a registrar service to track presence information. Thus, SIP-aware applications can not only gather diverse email and voicemail objects into one location but can also use presence information associated with individual users to provide "find me" and "follow me"style services. Such capabilities are especially useful for mobile workers because you can use SIP-based UM applications to present message data in a form most appropriate to the type of device with which a user is currently connected to the system. Microsoft is a big proponent of SIP and recently pushed through several extensions to the standard. SIP is the cornerstone of Microsoft Live Communications Server and is likely to take a more central role in the company's product lineup.
When it comes to actually implementing and managing UM solutions, your primary questions might revolve around issues such as storage, architecture, and performance. Most current Windows-based UM applications are closely integrated with Active Directory (AD). AD holds UM attributes, along with other Exchange mailbox properties, as user-object properties in the Global Catalog (GC) server. This approach offers a single point of administration for all user management, whether it involves conventional mailbox management or UM-specific configurations. The UM attributes can define voice-mailbox greetings, limits on received voicemails, or indeed whether voicemail is enabled or disabled for the user.
Storing such information in AD makes sense, but the first point of contact for an inbound call is the PBX, so some UM-attribute information must also be available on the PBX system. UM applications must provide some form of technology to synchronize the PBX user directory with AD user information. Most products typically use Lightweight Directory Access Protocol (LDAP) to implement this technology out-of-band from the IP telephony connection, as Figure 2 shows. The synchronization is generally one-way (from AD to the UM directory) but can be multidirectional in complex environments.
When a call is received at the PBX system and the caller wants to leave a voicemail message, the PBX system presents the call to the UM application (through the Exchange/PBX interface). The UM application accepts the call-setup request, then receives the data stream, which is in effect the stream of sounds that makes up the voicemail message. Whether the UM application is connected to the PBX via an IP telephony mechanism or over traditional E1 or T1 digital connections, the voice stream is already in digital format when it reaches the application. The majority of PBX systems are digital, as are the majority of Public Switched Telephone Network (PSTN) circuits, so they automatically digitally encode human voices upon receipt. In cases in which the PSTN trunk lines are analog, a digital signal processor (DSP) on the PBX converts the analog voice signal before relaying it to a handset or UM application. Digital voice signals are usually encoded into one of several possible formats, the most common being G.711, G.723, and G.729. (Javvin's online Protocol Dictionary at http://www.javvin.com/dictionary.html gives a good description of the G.7xx and other voice-encoding formats, casually referred to as codecs; technically, a codec—short for compressor/decompressor—is the software that implements a particular format.) Essentially, recent codecs are more advanced and have a lower bit rate—while maintaining reasonable quality—and thus consume fewer network resources during transmission. More important for a UM application, messages encoded in these codecs consume less disk space when written to users' mailboxes. Therefore, many UM applications will transcode a voice stream from a PBX in one format to a more efficient format (e.g., from G.711 at 64kbps to G.729 at 8kbps or GSM at 13kbps) at the expense of some quality.
What about the performance impact that running UM applications has on Exchange servers? Generally, adding a UM application to an Exchange mailbox server increases the system load associated with the mailbox by approximately 30 percent. Therefore a system capable of supporting 3000 Exchange mailboxes should be loaded with no more than approximately 2000 mailboxes when a UM application is present. This performance effect is caused by several factors, not least of which is the overhead of accepting calls from the PBX system and the computational overhead associated with transcoding the voice stream. Therefore, you might consider a UM system architecture similar to the one that Figure 3, page 4, shows, in which you don't install UM software directly on the same Exchange server that holds your mailboxes.
How much extra space will a mailbox need to receive UM voice messages? The answer varies depending on the number of voicemail messages the user is likely to receive, but assuming that the standard G.711 codec is used, a 30-second voicemail message requires 240KB of disk space. Using a more efficient but lossy codec such as G.729 requires only 30KB of space. So if your average user receives 10 voicemail messages per day, the storage requirements range anywhere from 300KB to 2.34MB.
Much like SMTP bridgehead servers, UM servers should be configured with adequate I/O subsystems. If the UM server is receiving hundreds or thousands of messages per day, these will all be written at least temporarily to disk during processing, so a fast I/O subsystem (ideally, RAID 0+1 with many spindles) is important. Similarly, sufficient storage capacity is a must, so that messages can be staged temporarily on the UM gateway system if the destination Exchange mailbox servers are unreachable. Most UM implementations use Messaging API (MAPI) to integrate into the Exchange environment, so getting messages into the Store and implementing onward transmission is actually quite simple.
Stored voicemail messages typically are represented in a .wav (or sometimes .asf) file format. Both formats essentially act as wrappers for a compressed audio file. Therefore, a G.729 encoded voicemail message can be played on the user's PC if he or she has software that can play .wav files and that has the G.729 codec installed. Accordingly, you might need to update your desktop environment by installing the correct software and codecs. Windows Media Player (WMP) offers such functionality; if a user tries to use the player to listen to a .wav file for which no codec is installed, WMP often automatically downloads the codec (as long as an Internet connection is available). Exchange 2000 Server and later also provide support for multimedia message formats (often referred to as m-cubed formats), and a multimedia control can be embedded within Outlook or Outlook Web Access (OWA) to render multimedia content within the context of the Read Message window.
But UM is not just about accessing voicemail messages from your PC. It's equally about accessing email functions from a wired or wireless phone. UM applications provide text-to-speech functionality that "reads" an email message to you while you're connected by telephone. Limited functionality is generally available for speech recognition as well, so that voice directives can be given over the telephone and parsed by the UM application. You don't typically see this functionality in today's Exchange UM solutions, but you'll likely find it in emerging products.
Fax services generally are handled in a similar fashion to voicemail services. Fax content is usually encoded in G3 or G4 format. (G3, aka the International Telecommunications Union Telecommunication Standardization Sector—ITU-T—T.4 encoding standard, is more common; G4, aka ITU-T T.6, is less common but can achieve higher compression ratios.) These formats are consistent with TIFF format files, so received fax messages are often represented as message attachments in TIFF format. UM systems generally use fax-modem technology for connecting with PBX systems.
Getting Behind UM
There are truly benefits to gain from moving to UM technology, especially if you host a large mobile workforce. (The Web-exclusive article "Messaging System Revitalizes Medical Center," November 2000, InstantDoc ID 16134, provides a case study of one organization's UM implementation.) But be sure you clearly understand the implications of deploying modern-day UM systems, which are increasingly integrated with real-time communications and presence-awareness technologies on top of traditional voicemail, email, and fax services. Thoroughly research the many UM products available; the Windows IT Pro Buyer's Guide article "Unified Messaging for Exchange Server," February 2001, InstantDoc ID 16451, and the Microsoft Exchange Solutions Directory at http://www.msd2d.com/exchange are good places to start. Web Table 1 (http:// www.windowsitpro.com/microsoftexchangeoutlook, InstantDoc ID 43902) also provides a brief summary of a few products that take a unique approach to UM integration.