IMS Q&A with Greg Papadapolous, CTO and Executive VP for Sun

Greg brings a diverse perspective that includes working on flight control systems for commercial airliners while at Honeywell, co-founding video conferencing pioneer PictureTel, and being Chief Architect at Thinking Machines, later acquired by Sun. He also was an associate professor of electrical engineering and computer science at M.I.T., where he conducted research in scalable systems, multithreaded/dataflow processor architecture, functional and declarative languages, and fault-tolerant computing. Greg envisions a truly networked world where every device can think or remember or sense or call out its name, thereby creating a system with the power to change society. The telecommunications industry represents a big piece of the puzzle and Greg offers his thoughts on its evolution.

In his current role, Papadapolous is responsible for managing Sun's technology direction and architecture, standards, the Science Office, global engineering architecture, and associated advanced development programs. He also provides leadership and consistency for hardware and software architectures across Sun. During his tenure with Sun, Papadopoulos has held several positions, including vice president of technology and advanced development for the company's systems business, chief scientist for server systems engineering and chief scientist for enterprise servers and storage.

We talked to Papadapolous about IP Multimedia Subsystem (IMS) and its impact on the telecommunications world today. Papadapolous explained that IMS takes us "from a world optimized for voice to a world optimized for data" giving users the ability to dynamically combine different types of services in a single call. He notes that today's network, while capable of delivering IMS type services, is not adequately equipped to do so. "The packet-switched domain has an interesting problem in delivering real-time multimedia services in that it can't guarantee quality level – it's like rolling the dice. IMS improves upon today's network by providing solid QoS so that users enjoy a consistent experience, Papadapolous says. He comments that "network convergence has accelerated the adoption of new services into an enterprise environment" and we can expect to see the lines between operators and enterprises continue to blur.

Sun's Identity Manager and Solaris 10 are two important technologies that provide significant value to an IMS infrastructure , Papadapolous explains, and he says that we can expect to see emerging technologies like JXTA play a role as peer-to-peer networks gain increasing acceptance.

For another perspective on IMS, see the IMS Q&A with Darrell Jordan-Smith, VP of Sun's Telecom Industry sales. Learn why the industry is excited about IMS, how customers can benefit from it, and why customers should look to Sun for IMS solutions.

What does IMS mean to you?
There aren't many things in the telecom industry that are more transformative than IP Multimedia Subsystem (IMS). IMS is network convergence at its best. We're rapidly moving from a world optimized for voice to a world optimized for data. In the world of IMS, you're able to dynamically co-mingle media types – voice , video, data – in a myriad of ways. This is what multimedia means – the simultaneous existence of several media types. It doesn't matter whether you want to talk, stream a video, instant message or email a picture, and by the way, do so simultaneously with two other callers. The network doesn't care and neither do you. It's just another data type. IMS is changing, irreversibly I suspect, the way in which we will communicate.

Why do we need IMS?
Consider the difference between the two worlds I just mentioned – one centered around voice, the other centered around data. Since the birth of the telephone, we've been communicating over circuit-switched technology. The current trend is to replace it with the more efficient packet-switched technology. Yet both provide the ability to connect to the Internet, send video or instant messages. So that begs the question: why do we need IMS if I can already do these things today? There are really three main reasons: QoS, billing and integration of different services. The packet-switched domain has an interesting problem in delivering real-time multimedia services in that it can't guarantee quality level – it's like rolling the dice. As a result, the quality of a VoIP conversation can vary quite a bit throughout its duration. IMS takes care of synchronizing session establishment with a QoS provision so that users have a consistent experience.

Billing is the second big issue. In today's network, all bits sent over the wire look the same. So how can an operator tell whether you're downloading a file, streaming a video or having a VoIP call? IMS provides the necessary information about the service being delivered to the end-user so they can appropriately bill you for it.

Finally, and in my view most significantly, IMS is based on Internet technologies and Internet protocols so that all services provided by the Internet are possible on an IMS network. This means that a multimedia session between two IMS users, between an IMS user and an Internet user, and between two Internet users is established using exactly the same protocol. This makes it trivial to do interesting things like setup a conference call with a video stream while sending a snapshot of a beautiful sunset to my wife on her computer. This is why I said that IMS is network convergence at its best; it uses the circuit-switched world to provide ubiquitous access and Internet technologies to provide compelling new services.

What are the key technological differences between today's telephone network and IMS?
One big piece of IMS is the notion of session. It's a way to provide call control and manage multimedia sessions over IP networks. Today when you make a phone call, a connection is established and torn down the instant you hang-up or attempt to do something else, like send an SMS or take a picture. There's no ability to combine these services together. A few years ago, the Third Generation Partnership Project (3GPP) decided to adopt the IETF-defined Session Initiation Protocol (SIP) as the basis of a new session-control layer for 3G core networks. Choosing SIP was a significant decision that ultimately is going to have manifold implications in the telecom world. With SIP, operators can can combine services from the circuit-switched and packet-switched domains in the same session, and for sessions to be dynamically created 'on the fly.' Things like adding a video stream while on a voice call will become commonplace. The beauty of SIP is that it's based on HTTP making it fairly easy to create new services. SIP service developers can use all the service frameworks developed for HTTP, such as CGI and Java servlets.

Another key area is mobility management. That is, the ability to find other users in the network and then establish a session with that user. The Call Session Control Function (CSCF) and the Home Subscriber Service (HSS) are the two key IMS components that enable mobility management. The CSCF is basically a proxy, which aids in the setup and management of sessions and forwards messages between IMS networks. The HSS is the other critical piece storing all the end-user data so that users can find each other over the network.

A third technological difference involves the way IMS controls the invocation of services and the interaction between the various service components. IMS is the uber-executor in the network if you will. Since IMS allows multimedia communication over multiple networks, there needs to be an efficient service provisioning functionality. IMS provides this by having the ability to identify which service or services need to be executed, the order in which they get delivered, and where to go find the services.

What part does user identity play in the IMS architecture?
User identity takes on a much greater role in an IMS environment and the HSS is at the center. It's helpful to think of the HSS as the nexus between the IT and the core network worlds. HSS acts as the central repository for user-related information such as security information (who am I?), location information (where am I calling from?), and user profile information (what services am I subscribed to?) to name a few. When you consider the 100s of millions of people with multiple profiles and multiple devices engaging in multimedia sessions with a huge array of interesting services this starts to add a lot of complexity to the system. IMS requires that your IT systems support complex subscriber identity structures, so that subscribers can have multiple "addresses." To do so, operators will need a robust identity management system that provides control and flexibility to take into account the key changes in the way that they manage and use subscriber identities.

How do 3rd party developers participate in an IMS world?
SIP is based on HTTP, arguably one of the most well known protocols in the Internet environment. Consequently, SIP service developers can use all the service frameworks developed for HTTP, such as CGI and Java servlets. This greatly simplifies the development of new services. IMS also provides a standardized architecture for enabling advanced IP service deployment. This means that IMS services can be developed independently and at the same time use the common features of the IMS infrastructure. When things are easy to build and deploy, you get growth and innovation. Look, there's already more than 1.4 billion people – about 1 in 5 of the world’s population – using mobile phones. Operators and other service providers are salivating at the chance to create new revenue streams from next-generation multimedia mobile services delivered over evolving IMS networks: multiparty games, video chat, ringtones, click-to-buy, "see-what-I-see" video sharing, location-based services - you name it. I expect to see a robust ISV ecosystem develop around new IMS-enabled services that will bring real excitement to communications.

Where does Sun fit into the IMS picture?
From the top to bottom – from the handset with J2ME down to the OS with Solaris on a Netra ATCA platform. We're all about systems engineering and IMS provides a great sandbox in which to play. IMS is an elegant manifestation that makes a highly complex and disparate network manageable. It changes the paradigm for communications, from stateless to statefull, from circuit-switched to packet-switched. There's lots of opportunity for companies that can provide the infrastructure that optimizes the efficiencies and provides the scale required for IMS. I spend a lot of time thinking about how to make network services run well for millions for end-users. Similarly, our products reflect an inherent design requirement to scale, to reduce complexity and to be reliable. For example, Identity management is going to be a huge issue in deploying IMS. Customers need to consider the impact of a complex identity structure that supports IPv6, a central prerequisite for IMS services. Our Identity solution is a great fit here since it was built to address the same complexity and scaling issues. It automates and integrates every aspect of identity management across wireless networks, including secure, federated provisioning, user profile management, federated identity management for single-sign-on and identity auditing to keep track of it all. You'll need a bullet-proof OS to run on and that's Solaris 10 – a proven operating environment in telco for the last 20 years. Finally, we offer a carrier-grade hardware platform at the bottom of the stack with our NetraTM ATCA product line – on SPARC or AMD Opteron.

For more information on Sun's Netra and upcoming ATCA products , go to http://www.sun.com/netra.

From your perspective, are the lines between operators and enterprises blurring?
Absolutely. Network convergence has accelerated the adoption of new services into an enterprise environment. Before the Internet, a business needed some phone lines and power to get up & running. Now, they need to provide access to corporate data, email, calendar and other tools to thousands of employees globally from multiple device types. Add to that the notion that not everyone has the same access right and privileges. Somehow the network needs to manage these users. Let's take Sun as an example. Much of the work force is "virtual" utilizing a multitude of service provider and access methods. Our CIO, Bill Vass, has provided a global network that provides access to all my data from any Sun Ray when I insert my Java Card, or the equivalent of "channel Greg." The network authenticates me and I'm now looking at my desktop as I last left it. In this example, Sun authenticates me while the operator provides the bandwidth. But I might just as easily be authenticated on a operator's network in a hosted environment. In fact, that's likely to happen as I access the network from another personal device. My guess is that this trend will evolve as individuals and corporations take on multiple "personas" when they enter the network. In the morning you might be a sports enthusiast, during the day an employee, on the weekend a member of a charitable organization and so on. The question then becomes who determines your access rights, privileges and to some extent, billing scenarios? Increasingly, I suspect we'll see the lines of demarcation between operators and enterprises fade as the network evolves.

Where do technologies like JXTA potentially fit in the IMS landscape?
Peer-to-peer (P2P) networks like JXTA will enable the Internet to reach the next level of security, management and scalability required to support the proliferation on information exchanges between peers and the plethora of new client devices. In a converged network where multimedia sessions are created, you can envision some compelling services that have applications of P2P to SIP in particular, and Internet communications in general. It's easy to envision P2P conferencing, real-time video streaming or broadcasting using JXTA as the underlying P2P technology. This could unleash an explosion in new forms of communications and create opportunities for innovative operators to deliver revenue-generating consumer and business services. The issue at hand, however, is the lack of standards around for SIP for the peer-to-peer environment. The good news is that a few proposals have already been submitted to the IETF to consider extending the traditional SIP infrastructure to include support for P2P networks.

For more information on JXTA, go to http://java.sun.com/.