"Presence Everywhere" is the mantra for an increasing number of unified communications vendors and providers. As we've explored in previous articles, presence information is a powerful trigger for starting a wide range of conversations, whether they are voice, video, text, group chat, social networking, whiteboarding, or some other form of collaboration.

When most applications and devices are presence-enabled, your core presence engine needs to scale in a serious way. And when you need to squeeze every ounce of performance out of your presence engine, the underlying choice of technology matters a lot.

Choosing a presence technology involves many factors, even with regard to the dimension of scalability. One such factor is the wire protocol used for communication between the presence engine and the myriad devices and applications that produce and consume presence information. Naturally you could create your own wire protocol, but to avoid vendor lock-in and ensure interoperability with current and future presence-enabled devices and applications it makes sense to use one of the standardized presence protocols.

There's an old joke among technology architects: "The great thing about standards is that there are so many to choose from".

Unfortunately, that's true in the presence domain as well. The primary technologies here have both been developed under the auspices of the Internet Engineering Task Force (IETF), which over the years has produced most of the core technologies that power email, Voice over Internet Protocol (VoIP), instant messaging (IM), and the web.

The two IETF presence technologies of interest are SIMPLE and XMPP. SIMPLE is short for "Session Initiation Protocol (SIP) for Instant Messaging and Presence Leveraging Extensions" and represents an effort, ongoing since 2001, to extend the SIP multimedia negotiation protocol to support presence and IM. XMPP is short for "Extensible Messaging and Presence Protocol" and represents IETF formalization of the presence and IM technology originally developed in the Jabber open-source community back in 1999.

Both SIMPLE and XMPP are being widely deployed, so it should come as no surprise that technologists are beginning to analyze their performance characteristics, and to develop optimizations where appropriate.

In particular, work is happening within the IETF to define a common model that will enable objective comparison of presence technologies for various use cases (see draft-ietf-simpleinterdomain- scaling-analysis and draft-saintandre-xmpp-presenceanalysis). Of special interest to date has been bandwidth usage, which pales in comparison to what is required for voice and video but is of interest nonetheless, especially in very large network peering arrangements and constrained devices such as mobile phones and tablet computers.

The results so far indicate that XMPP is perhaps 10 times more efficient than SIMPLE in terms of bandwidth usage. There are three main reasons for the disparity:

1. SIP presence subscriptions are short-lived and must be refreshed periodically (by default, once per hour), whereas XMPP subscriptions are long-lived.
2. All SIP packets are acknowledged, whereas XMPP presence packets are not typically acknowledged.
3. SIP presence packets are about five times larger than XMPP presence packets.

Naturally, the technologists involved are not sitting still. The SIMPLE Working Group has developed mechanisms for sending partial notifications, thus cutting the packet size, and for sending updates only if the presence state has changed, thus obviating the need for many notifications. The XMPP Standards Foundation has defined mechanisms for offloading rich presence such as user location, mood, and the like from the core presence transport onto a generic publish-subscribe transport called personal eventing, and has also defined ways to use standard compression algorithms to cut down on bandwidth usage. And you can be sure that further optimizations will be forthcoming from both groups.

There are plenty of business and technical reasons to choose either SIMPLE or XMPP, and all the smart presence vendors support both so that they can satisfy a wide range of customers. The choice is not necessarily "either-or" but often "both-and" even inside the same organization. That said, technology architects need to be finely attuned to the implications of their design decisions, so they may want to pay attention to the ongoing analysis of presence scalability, as well as real-world experience with these technologies on the never-ending experiment we call the Internet.

Achieving Internet-grade scalability is a difficult task, and we don't yet have proof that any existing presence technology can be scaled up to such heights. Yet we know that as the presence revolution rolls along, unified communications vendors and providers will increasingly build and deploy real-time applications that use presence as a kind of Internet dial tone. As a result, the core presence engines powering that revolution will be challenged along the way. But who ever said that changing the world or just your organization was easy?

Joe Hildebrand is CTO of Jabber, Inc. Telecom experts know Jabber for its protocol known as the Extensible Messaging and Presence Protocol (XMPP), the first protocol to be ratified by the Internet Engineering Task Force (IETF) as a standard for presence and messaging technologies. For more information, please visit the company online at www.jabber.com.

Unified Communications Communications Magazine Table of Contents