Market Analysis: Unified Messaging

We've said it before, but low UM adoption numbers suggest some organizations aren't getting the message: You need real-time communications systems that connect your users with the people they must talk to, while routing less urgent matters to the message repository of choice. Our testing (see "A La Carte Messaging") puts to rest the theory that these projects are too big and complex. The Siemens offering we tested starts at a measly $80 per user, per year. If you can't afford that for your multimessaging salesforce or high-powered executives, it's time to rethink your communications strategy--your company probably pays more for monthly pizza lunches. SIP phones are coming down in price as well, and UM systems will integrate with and leverage network resources. It's not really new technology; it's making the most of what you have.

Don't wait until the next generation of knowledge workers hits the enterprise. The killer application, e-mail, will be ancient history for them. They will give it back to you and start using their cell phones, IM and SMS or text messaging. What then?

Your Competitors Get It

Our reader poll for this article painted an interesting picture of UM adoption. Only 17 percent of 686 respondents use UM. Forrester's latest report on UM adoption parallels our findings. But that's not the end of the story: Every one of our respondents from organizations with more than 5,000 employees has implemented UM. Is this because large organizations are more complex, with multiformat messaging needs, and they've found UM boosts employee efficiency? Or is it because only large enterprises can afford a complete UM system, with attendant network upgrades and new phones?Both. Having a single point of access was the No. 1 benefit of UM cited in our poll. As for cost, even companies that would benefit greatly from UM may not have the resources to go whole hog. Many vendors let you start small, with voicemail, for example, then gradually add e-mail, fax and specialized UC features, such as FMFM. In addition, as open standards continue to infiltrate proprietary systems, the cost of UM is falling. Standards also improve integration with other enterprise systems, for example, with common directories (such as Microsoft Active Directory and SunOne iPlanet) and e-mail message stores (including Lotus Domino and Microsoft Exchange).

UM isn't for every user, of course, no matter how much vendors want you to think it is. Employees who interact almost exclusively in one medium, whether phone, e-mail or face-to-face, won't see many benefits. Still, in the next 18 months, 30 percent of poll respondents who don't have UM will be investigating it. That number corresponds to UM revenue increases forecast by analyst firms year over year (see chart).

Security Concerns

VoIP (voice over IP) applications incur the same risks as other IP services--a good example is the data-over-IP problem of DoS (denial-of-service) attacks. The good news is that the same protections you use for data-- firewalls, antivirus, IDS--apply for voice. But there's more to it.For SIP-supported VoIP implementations, SIP can set up sessions and prevent third parties from hijacking a call. And SIP phones can work behind a firewall with STUN (Simple Traversal of UDP through NAT). STUN enables a SIP phone to use an external server, such as a registration server, to route calls to and from the SIP endpoint. Addressing calls through a registration server or a session server means that your SIP phones are not generally available to the world, only to other phones that register and authenticate with the same server.

In addition to data security, privacy should be a top concern. If you are a federal government entity, privacy safeguards are required if your voice is transmitting classified data or sensitive information. IP phones are not tethered to the PSTN, where tapping needs a physical connection to the wire. If calls are sent over the LAN, WAN or Internet, anyone with a protocol analyzer can tap the call simply by capturing and analyzing the voice packets. Fortunately, VoIP systems include encryption (see "Keeping IP Voice Safe and Sound, and "The Real State of VoIP Security"). Make sure encryption is implemented right down to the IP phones.

Finally, if your IP PBX runs on Windows, as those from Interactive Intelligence and Siemens do, make sure you secure it with firewalls and antivirus. Note that Interactive Intelligence's Communité comes with an McAfee antivirus application, and 3Com implements its own firewall based on GNU Linux's ipTables.

Although UM products do put varied messages into a universal inbox that users can manage from an e-mail client, Web browser or TUI (telephone user interface), they don't provide all the niceties we'd like--for example, speech-to-text translations to read voicemail, and document conversion for incoming fax to text. We did see some nice voicemail implementations, though, as well as fax servers. Those who already run fax servers need only move fax messages into the e-mail message store, where they are manipulated like other message formats.

Current UM voicemail services integrate with a PBX or central office switch to record incoming voice messages and store them for playback (see "PBX Outlook"). UM voice-mail systems are generally priced per port, based on the number of simultaneous messages read or recorded. The goal is for the PBX to forward voicemail actions and storage to the UM system.Voice messages transferred to the UM voicemail system can be accessed directly from the UM platform as voicemail. From there, they are copied to a IMAP4-compliant universal inbox, such as Microsoft Exchange or Lotus Domino. UM systems convert voicemail to digital .wav sound files and deliver them to your message store. UM systems that provide dual message stores also can synchronize an IMAP4-compliant inbox to provide some redundancy should the e-mail server go down. But note that messages usually aren't synchronized between the e-mail server and the UM store with voicemail. In other words, accessing the UM system as a backup may show voice messages long deleted from an e-mail store, depending on the configuration to purge old voicemails from the UM platform.

Because the voicemail sent to the message store consists of an attachment to a message file, users can open the attachment to play the message from the default media player on their PCs. They also can listen to voicemail, e-mail and even fax message headers from the telephone interface. And with Interactive Intelligence's Communité, users can even listen to their calendar entries while commuting to the office, thus saving time, the raison d'être of UM.

The fax server may be an integrated service or module on the UM platform, or current UM software can support third-party fax servers like Captaris' RightFax or Omtool's Genifax. As long as your server can deliver fax messages to the e-mail store, you should be good to go with any UM platform. Users will be able to read faxes in e-mail and listen by phone to the header messages as the header string displays across the top of the fax page. This info will, at the least, include the sender's number. It also may include sender name, recipient name and number, and some information on the fax itself. However, employees will not be able to forward to another fax server unless the message in the inbox is classified as a fax by the e-mail system. That capability would come only from a fax server that provides a fax connector, such as those from GFI Software and Biscom.

Current UM platforms also include TTS (text-to-speech) engines to translate e-mails and headers to voicemail and fax messages. In most cases, third-party software from Microsoft or ScanSoft will be included in the UM package. Improvements in speech recognition and TTS have greatly benefited UM. TTS engines make UM more efficient--we found speech translated to text at two to three times the speeds possible in previous years. What you still won't find today, however, is STT (speech-to-text) engines. Two of three participants in our tests let you give simple commands, like "delete," while interfacing with an inbox through a phone system--a very limited STT application. We'll be watching for STT to catch up with TTS. When STT matures, UM will be even more efficient--most users can read an e-mail message much faster than they can listen to it play back.

Until then, we'll have to be satisfied with our new buzzword: UC. We'd add to our definition of UC by saying it's everything that happens to a message in real time, before it goes into a message store for replay. This includes presence management as well as call routing rules for FMFM. UC features let a caller follow and find a call recipient based on real-time routes they place in their user configurations.Another welcome addition to UM customization. People are getting used to setting up their cell phones with special ring tones and assigned rings, for example, and they are bringing those habits to work. Current UM platforms let users prioritize how voice, e-mail and fax messages are read over the phone, or set online presence availability as "available" or "at a meeting."

UC is a step in the right direction for UM because it takes UM beyond integrating with an IP PBX to obtain and replay voice messages. When a call comes into the UM platform, it may route the call back to the IP PBX, if a call recipient has configured his or her FMFM to call another phone, before the caller is shuttled into voicemail. This is called a SIP referral; its success depends on the SIP standard being in place right down to the IP phone, and this brings into play interoperability issues between IP PBXs and phones.

Many vendors, including Avaya and Mitel Networks, have added SIP support to their PBXs and UM platforms to allow voice, video and data over a common signaling (SIP) and transport (IP) protocol. SIP-based servers scale easily and can provide additional services, like IM, FMFM and even Web conferencing.

SIP-based PBXs have wide appeal for both small and large enterprises. For small companies, a SIP PBX is an ideal upgrade from a key system (a premises-based phone system where the telco handles PBX functions). It's affordable, provides some continuity in features and sets the stage to integrate productivity apps. Vonexus Enterprise Interaction Center and Zultys Technologies MX250, for example, provide Windows-certified IP PBXs for small and large companies. For enterprises, Siemens' HiPath 8000 system scales to more than 100,000 users from a single location. This will help consolidate telephony build-outs in multiple locations and bring them home to a central data center to reduce operating costs. But pay attention to the hardware compatibility lists provided by your vendors for SIP: If you don't use VoIP, you'll need an analog telephony board from the likes of Dialogic.

When it comes to signaling between IP PBXs and phones, the market is full of incompatible proprietary codecs: Cisco is focused on Skinny, 3Com has its H3, and Mitel squares off with MiNet. But SIP has captured the hearts and minds of developers at Avaya, Interactive Intelligence, Siemens, Zultys Technologies and others--when you can give a phone intelligence (via SIP), that excites users and creates new products.But there are drawbacks. SIP is a standard peer-to-peer protocol, where the intelligence resides in the endpoint. However, that intelligence is not as sophisticated as we'd like in terms of passing the business-class features driven by IP PBXs down to proprietary and supported third-party phones. One example is SIP's lack of a shared-line appearance indicating to one phone that another is in use or off the hook.

The SIP standard is young, and it has not supplied all the functional definitions and call flows to replace feature sets found on digital business sets connected to TDM-based PBXs or even a single vendor's IP phones and PBXs. Also, every vendor interprets SIP differently, especially because the standard deals with some 50 call flows, and enterprises are looking to replace their PBXs and full-featured digital phone sets that have hundreds of features users don't know how to use.

Technically, implementations may fit the SIP standard, but they may not interoperate with other devices, such as phones and gateways. Vendors interpret the standard for items not spelled out, such as displaying multiple line appearances, bridging, conferencing, camp-on and distinctive ringing. Even handing off calls from a SIP server like Asterisk to two clients and forwarding the call can be problematic, depending on the IP phone.

When you start discussing IP phones, another variable enters the fray: Not all SIP/IP phones are equal (for comparisons, see "Session Initiation Protocol Phones: The Low-Cost SIP Phone Challenge"). When you set out to mix and match IP phones, first determine the features you must support.

SIP-based UM products are making it more affordable to handle multiple message types. If you must replace a PBX this year, investigate SIP for an upgrade. Staying with one vendor will obviate testing for interoperability. And in the coming years, you'll have the option to buy SIP/IP phones or gateways from third parties as the standard matures.SIP is a work in progress. But it's good work and holds the promise of a standards-based IP telephony infrastructure, especially with the forthcoming SIP B, a working agreement by multiple vendors such as Polycom and Siemens (though not an IETF standard) to implement advanced feature sets on phones in a similar way. Once these feature sets become standardized, enterprises will be able to assemble best-of-breed systems at ever-lower costs.

Most people believe e-mail is the killer application for small and large enterprises. The truth is, e-mail is simply the next best thing to real-time communications. The real killer application is the one that can package and deliver voice in real-time, such as a PBX.

The reason no one has identified the PBX as a killer application is because it's always up and available, routing calls to desktops, auto-attendants, IVR (interactive voice response) systems and voicemail. PBX platforms have taken uptime to new lengths and have always been a good deal for enterprises.

To date, TDM-based PBXs with proprietary phones have dominated the enterprise market. These systems come with IVR modules and UM (unified messaging) components that are tightly integrated on the PBX system itself or interconnected using serialized line cards. From an integration point of view, enterprises went with one vendor to obviate integrating with third parties.

The IP PBX has emerged as an alternative to the TDM-based PBX and is now poised as the next killer application. IP PBXs operate on a LAN using packet-switching technology rather than a circuit-switched technology. Voice calls traverse the LAN and become another business application, albeit one requiring HA (high availability) and QoS (quality of service) schemes. Enterprises can mix and match phones from different vendors depending on the feature sets required on the desktop (see "IP PBXs: Live in Our Labs"). Furthermore, IP phones are easier to install and move than conventional phones. An IP phone can be shifted to another port without programming or a great deal of administrative support, and features can be configured from a PC using a Web browser.In 2004, unit sales for conventional PBXs declined due to their replacement with IP PBXs. The outlook for these conventional PBXs will continue to darken, to fewer than 3,000 units shipped by 2008, according the Telecommunication Industry Association, even as the market for IP PBXs accelerates. And that acceleration means reduced prices and increased functionality.

Do large companies know something their smaller rivals don't? It appears so from our analysis of who is deploying unified messaging. Although overall adoption is under 20 percent among readers we polled, every respondent from a company with more than 5,000 employees has UM. In "Messages Unite!" we explore the UM market--what's new in features, security and standards.

For our product review, "A La Carte Messaging" we started with the premise that SIP (Session Initiation Protocol) was a standard, and each participating vendor should be able to use a SIP-supported IP PBX, specifically the open-source Asterisk. The promise in the premise is that enterprises can reduce the cost of a PBX down to its heart--an intelligent switch--then put saved resources into a UM product with voicemail.

But the more we talked to vendors, the more discouraged we became. Most would not play with Asterisk. From their business-model point of view, customers want fully integrated and interoperable (read: single vendor) IP PBXs, voicemail systems, IP phones and IVRs (interactive voice response). Many also told us it's too soon for enterprise-class SIP-based UM. Maybe, but we think that day is imminent, and it makes good sense to start planning.

Interactive Intelligence, Siemens and 3Com sent their UM offerings to our Syracuse University Real-World Labs®. We had some interoperability woes with Asterisk, but on the whole we were impressed by all three products. Interactive Intelligence's Communité squeaked past Siemens' HiPath Xpressions for the win. We commend each vendor for its commitment to SIP and would consider using any of these offerings.