|Hype-Busting: IMS Is Gonna Take a While|
By Edward J. Finegold - October, 2005
Billing World & OSS Today
The telecommunications industry seems slow to learn certain lessons, like those that demonstrate the dangers of over-hyping innovation. Once again the industry finds itself predicting an ambitious future, often without citing the massive hurdles that must be overcome. The acronym IMS—which can stand for either IP Multimedia Subsystem, or IP Multimedia Services, depending on who’s talking—is being tossed around the industry with little regard for what it really means and what promises it is intended to deliver. The dirty little secret, however, is that implementing true IMS architectures is a long way off.
In the more immediate future, for most large U.S.-based telcos, is nothing overwhelmingly innovative—VoIP, maybe even SIP-based VoIP and television, however it might be delivered. VoIP and TV may be new to the telcos, but they aren’t new to customers. Hence, the question arises: When will we see telcos deliver something truly new, based on real IMS capabilities? In short, there are some things that look a little like IMS, but true IMS services—the kind of things marketers and trade show speakers love to hype—are several years away.
What Is IMS?
To set the record straight, IMS means IP Multimedia Subsystem. It represents an organizing principle and architecture for building cross-domain service interworking in an all-IP environment. “IMS is a facilitating framework,” explains Roberta Cohen, vice president for IMS business development at Telcordia Technologies. “It allows carriers to deliver converged services, … but these are two different things. One is a facilitating architecture, the other a set of services. You could offer these new services—like VoIP, IPTV and fixed-mobile convergence—but without IMS it wouldn’t be as easy to interwork these things. No one has really implemented IMS yet.”
Others agree with Cohen’s assessment. “Customers are still going through their education around IMS, what it is, and what it enables,” says Mark Nicholson, CTO at Syndesis. “I see IMS as a next-generation control plane.” Nicholson compares the development of IMS to that of SS7 and the Intelligent Network (IN). “We had in-band signaling, but when SS7 was invented it allowed us to separate control from the voice path,” he says. “IMS is doing the same—bringing in a new control plane for the IP network that gives us a separate control layer to enable greater service interactivity and control over that interactivity.”
Nicholson also points out, however, that IMS architecture makes certain assumptions that have not necessarily been realized today. For example, IMS assumes an underlying IP and connectivity network with all transport and access aligned to support all-IP capabilities from premises to premises. Once the all-IP network layer is in place, IMS can begin to work its magic. Until that time, however, IMS will be limited by the availability of IP networking.
So IMS is an enabling architecture, and IP services are IP services. The two do not necessarily need to co-exist, though most carriers are beginning to build out IMS operations architecture with future IP service interworking in mind. Though most VoIP devices being installed today are IMS compliant, IPTV launches that carriers like SBC and Telus have scheduled for the coming months, as well as upcoming content services, are not based in the IMS domain. For now, these services are being rolled out on a one-off basis.
“Most content is not delivered over IMS,” says Chris Daniel, senior director of business development at Leapstone Systems. “Could you do IPTV over IMS? Sure, if you put the energy into it. Could you do gaming and streaming? Probably, but I’m not sure that’s going to happen. In the next five years, you’ll have those three domains independent but interconnected.” As a result, services will come to market that look like they utilize IMS interworking, and may be delivered over a common pipe like a residential DSL connection, but an IMS architecture implementation will not yet be in the background. (See sidebar, “BT and IMS-like Services.”)
What’s the Big Problem?
The barrier separating today’s environment from the reality of IMS is in fact a set of complex problems, some of which telcos are just discovering, and others they’ve been combating for several years. “In my 10 years with a service provider, I’ve learned that with these kinds of projects, half of what you’re doing is figuring out what doesn’t work. That’s no one’s fault in particular, you’re just dealing with a lot of moving parts that are new,” says Daniel.
The first difficulty is that, because IMS is so new, the market suffers a general shortage of carrier-class IMS components. This is a typical problem for any new technology set, and it’s somewhat safe to assume this will be solved over time, likely in the near future. Carriers are pressing suppliers—those that build network elements and application servers —to fall in line with IMS. Some carriers are taking a hard line, telling suppliers “you need to evolve your product to fit into the IMS architecture. If you can’t commit to that for initial deployment, you’re no longer a player,” says Daniel.
Stories abound of equipment manufacturers struggling to deliver not only systems that support IMS, but also solid Management Information Base (MIB) databases that make their gear manageable. A source inside BellSouth who has spent the past several months working on the LEC’s VoIP architecture says that major equipment manufacturer has lost an opportunity to supply application servers based on the lack of reasonable MIBs. If this is the type of basic problem holding up IMS development, it’s clear that most IMS gear is simply not ready to come out of the oven yet. “We’re still seeing 24-month [estimates] before we see widespread deployment of standard systems,” says Syndesis’ Nicholson.
Second, as mentioned, to realize the full potential of IMS, an all-IP network environment must be in place, which is not the case today. IP networks are extensive and growing, but it is not reasonable to suggest that all telco customers are connected to a core IP network over a common IP pipe. Residential broadband in the United States is still in the minority, and most enterprise customers have not migrated their range of access services to a single IP pipe, though they are calling for native IP and Ethernet services from carriers. As IP networks are being built out, remember that most of the technology involved—such as MPLS—is also still very new, which means the problems in implementing these new network technologies are still to be ironed out. (See “Managing Capacity in IP,” Billing World and OSS Today, September 2005.)
Looking deeper, IP developments are happening on top of changing layer 1 and 2 transport and access networks. The optical layer is a mix of SONET, WDM, DWDM, PON and other technologies that are being engineered to coexist. Layer 2 networks are gradually shifting from ATM and frame relay to various flavors of Ethernet. In short, there is significant instability in these underlying networks. Before there can be a dynamic, all-IP, IMS-based environment, layers 1 and 2 must stabilize to provide a solid foundation to support the real-time capacity, quality and service delivery capabilities the IMS control plane assumes will be in place.
Lastly, as these networks drive toward the all-IP future, operational organizations remain in flux. Most major carriers are still consolidating their operations in order to attain a centralized, service-centric view of their entire network base with efforts to roll all ordering, provisioning and diagnosis into common streams. In addition, this is all happening in the context of recent mega-mergers, which further complicate these efforts. Carriers are banging away at this problem to try, for example, to automate provisioning from layer 1 up through the services layer, at least in transport networks (most access pipes will continue to be provisioned in a linear, mechanistic fashion for the time being). In short, underlying networks are not yet aligned or managed in such a way as to allow them to meet the needs of a high-volume, dynamic, real-time services environment. Until the underlying network and operations problems are solved, the service examples being hyped today will not be realized.
What’s at Issue Right Now?
Those big picture issues are going to take years to solve. In the meantime, the folks in the trenches are tackling these problems incrementally, but issues from carrier to carrier are not necessarily the same. “I don’t think there’s a lot of commonality,” says David Sharpley, vice president of marketing and product management at MetaSolv Software. “We see delays because of hardware and software not being ready, because of changing business requirements, and a lack of understanding of the impacts on operational processes and the people who use them. You’re into human change now, and that’s a big challenge for large companies. We haven’t seen a common thread, though, because each project is so unique.”
Certainly, every carrier has unique technical, organizational and personnel matters to overcome, and the biggest challenges aren’t necessarily centered on IMS yet. For example, companies like Telus and Telecom Italia are focused on delivering broadband, unified IP networks, VoIP and TV while also consolidating their operations. SBC’s Project Lightspeed operations architecture is being built from the ground up to support its triple-play launch and future IMS offerings. The approaches are different, but what’s common is that they are still wrestling with OSS problems that have more to do with operating underlying networks today than with delivering an IMS architecture.
“Putting the service out there isn’t the problem,” says Telcordia’s Cohen. “It’s how to cost-effectively operate it when you scale it out to a large audience. Carriers are getting sticker shock over the operational cost of going to market quickly. The cost is so high because so many systems have to be touched and revamped.”
Provisioning Turned Inside Out
Provisioning is one of those systems being overhauled, and in fact turned on its head. Until now, telco provisioning has been almost entirely linear, mechanistic and silo-oriented. A complex order is broken down, provisioning orders are sent to each relevant network silo, and a linear, mechanized process is followed. If something goes wrong, the order is excepted, thrown back into the queue, and the process begins again until everything is delivered. Services are often tested after the fact, and as any telco customer knows, the entire service provisioning process can take weeks or months to complete. This model is becoming irrelevant.
In the new model, the IP environment is supposed to be about rapid service introduction, on-demand capability and ubiquitous access to services. The network must be able to provide the proper connectivity from the point of origin—like an application server—to the customer device. To deal with all of the complexity, what needs to materialize in the IP and IMS domains is the ability to create standard service components that can be mixed and matched to build targeted service offerings very quickly. These service components will include things like network connections, IP streams, applications and features, as well as capabilities such as presence, QoS management and various types of data that either relate to service performance or must be delivered to the user.
These service components will ideally be managed in a way that feels like web services integration: different components will announce their capabilities, and some central, organizing system will be the source for calling on these capabilities to tie them together. The ability to mix and match reusable service components is definitely an element of the IMS domain and architecture. Companies like Leapstone are supplying just the first IMS building blocks to SBC and others today. Admittedly, little of this functionality is automated, and most service components must be defined manually in the IMS applications designed to handle service creation. That said, it’s an incremental process that will require several iterations to perfect.
BT Embraces Service Components
BT sees this service component approach as they key to managing the complexity of personalized, targeted IP services. “You do have a significant clash between the different ways you operate these different networks,” says Daryl Dunbar, BT’s director of 21st Century Network (21CN) design and development. “We’re borrowing from the software development model … ideas like objects and reusable components. To accelerate service introduction, we want to let people pick and mix components to build their own services.”
The idea is to define reusable service components and assign systems and network “hooks” that reveal them as reusable components. What’s defined is all of the capabilities of the service and its network and provisioning requirements. These components can then be grouped to build specific services that are ready for delivery. BT plans to make these hooks available internally and to wholesale customers to enable them to combine BT’s capabilities with their own to create new services. This is perhaps the initial inkling of the MVNO model being realized in the landline, IP realm.
“I can … build a product much more rapidly than ever before,” says Dunbar. “We’re targeting dropping the service introduction time down to 9 weeks.” With that in mind, Dunbar also says that access provisioning isn’t going to change radically from the traditional process, but “once you’re connected to the cloud, the service you want is dynamic.” In other words, facility provisioning may not change much in the short term, though many carriers are looking at concepts such as “left in place” networking and pre-building access pipes into new construction that can provide access to service provisioning the day a customer moves in. Service provisioning, however, is headed toward a real-time mode.
Device Support a Vendor Issue
Another major challenge in provisioning has to do with the number of new devices being introduced to the network and service layers. “There’s constant change in hardware suppliers but also in the number of devices provided by each and how unique each one is, plus the different IOS and software versions on each device,” says MetaSolv’s Sharpley. Carriers are telling OSS vendors to collaborate with equipment manufacturers to ensure device support, but often the carrier needs to press the equipment vendor to answer an OSS vendor’s calls. Those with established relationships and protocol for joint solution work have the advantage, because the message coming from many carrier CTOs is “make sure you’re there when we are.” At a recent TeleStrategies conference, a spokesman from BT made it clear that it considers device support the responsibility of the network equipment manufacturer, hence placing the onus on NEMs and OSS vendors to collaborate.
Telus Presses Vendors
BT is not the only carrier to expect more proactive collaboration from its vendors. Despite well publicized labor issues, Telus is aiming for an extremely aggressive IPTV launch which, at press time, was slated for the September 2005 timeframe but not yet announced. According to vendor business developers with direct access to CTO-level staff, Telus has laid the burden on its OSS and network vendors to take on the cost and risk of delivering joint solutions to meet its deadlines, while insisting on a high level of software engineering quality.
In exchange for meeting these demands, Telus is making large, long-term commitments to six key OSS/BSS suppliers, though not all of them will necessarily stay in the game, depending on their performance. The current list includes Amdocs, NetCracker, Syndesis, Intelliden, Atreus Systems and Micromuse, and each is responsible for collaborating collectively where necessary and with network vendors Alcatel and Nokia to support ADSL2+, IPTV, VoIP and future offerings.
As Telus presses forward, Telecom Italia and Telefonica are taking different approaches to set the triple-play pace in continental Europe.
Telecom Italia was one of the first to realize it needed to centralize and streamline its operations to achieve a seamless, pan-European IP and broadband network, and its foresight is a big part of the reason the company is able to win DSL business away from major PTTs outside Italy. Even with that head start, TI is just beginning to roll out basic IPTV offerings in limited trials and is being very careful about its service quality in the process. The company is running into issues like loop quality, choppy QoS in the service stream, and the fallout from integrating brand new vendor software and network gear. Its goal is to maintain a strong customer experience when it launches network-wide, by first ironing out the majority of these technical issues.
Other European carriers, such as Telefonica, are moving into IPTV more rapidly and with less of a focus on solving quality problems up front. Telefonica has already gained more than 40,000 IPTV customers and has deployed more than 100,000 IPTV ports in Lucent DSL access concentrators. At the same time, Telefonica is has been migrating its transport networks to an all-Ethernet architecture for two years and plans to merge its commercial and residential Ethernet networks by the end of 2005.
Technical staff members and supplier business developers say that Telefonica’s subscriber numbers have been achieved with a significant customer churn relating to dissatisfaction with the initial quality of the 50-channel television and video-on-demand service. This reflects the immaturity of the technology and Telefonica’s willingness to keep marching and work out the kinks along the way. Telecom Italia and Telefonica are worth watching for companies like SBC, to determine whether customers will accept poor quality initially and continue to try out new services, or instead whether carriers have only one chance to make a first impression when it comes to TV.
Today, SBC’s video service is the center of focus and the source of some of its biggest challenges. IMS is in the works, but is not center stage at this point. It’s basically common knowledge that Microsoft is dealing with a number of problems around IPTV. Senior business developers close to Lightspeed’s executive managers and chief architects report that executives at SBC were shocked when confronted with the sheer number of servers necessary for Microsoft’s IPTV solution and are therefore concerned with its scalability and related costs. Microsoft may be the only player with the resources necessary to overcome this problem, but no one knows when it will be solved and therefore ready for mass production.
SBC is also reportedly very concerned about quality, particularly in the home. Currently SBC has no technology to provide the reach or visibility necessary to manage independent service streams delivered over a common pipe and distributed to various devices in the household. It can test loop quality prior to service initiation, but once the service is up and running, there’s virtually no visibility down to the set-top.
For that reason, the carrier is issuing RFXs for agent technologies that can provide quality measurements and problem diagnosis. Metrics such as channel change response time and video quality are high on the list—especially considering that it can take several seconds to change channels in some IPTV implementations. In the VoIP domain, SBC wants to see SIP call setup metrics and voice quality measurements. Altogether, one major goal is to understand how one service affects the quality of all the others traveling over the same access pipe. The carrier is also looking at ways to build capabilities such as virus and worm detection directly into the pipe. BellSouth reportedly shares these concerns as it examines its VoIP and video offerings.
Industry rumors have it that SBC plans a “controlled market entry” in one market around the end of 2005, with plans to expand the offering to enter new markets in mid-2006 which makes a lot of sense. It is believed that the company has begun to install its IP video equipment and is building out its IP video operations center to monitor the network, a super-hub to acquire national video content, and four video hub offices for storage and on-demand delivery to subscribers. The operations center is scheduled for full operation by year end, with the super hub and video hubs operational in 2006 and 2007. Regardless of these plans, it’s likely that the initial TV offering will not be par with cable offerings, as challenges such as multi-second delays when changing channels persist.
Sources with a major systems integrator responsible for delivering significant portions of the architecture confirm that SBC’s initial Project Lightspeed trial is slated to launch in San Antonio, Milwaukee and Kansas City in November 2005. This was initially the Lightspeed launch date, but it has since been re-categorized as a trial. The launch is now slated for April or May 2006 and will include IPTV, Internet access and VoIP, but not video on demand—in other words, basic triple play. An April-May launch would match the timing of SBC’s second planned OSS/BSS upgrade.
Currently, OSS and IMS vendors involved in Lightspeed include Amdocs, Granite, SAP, Syndesis, EMC Smarts, Micromuse and a variation of Trendium’s solution. IBM and MQ Sonic are providing integration middleware, and a joint IBM-Leapstone solution will serve as the IP service delivery platform—an IMS component. Network providers include Alcatel, 2Wire and Microsoft for set-top boxes. In the midst of these triple-play launches, SBC will migrate its VoIP service from the AT&T CallVantage platform to one being developed internally, though reasons for the switch are unclear. Whatever the cause, it’s clear that SBC is dealing with a large number of new moving parts, making the delays in what were optimistic launch schedules not very surprising.
When We Build It, Will They Buy?
Ultimately, the point of overcoming all of this outrageous complexity is to increase the subscriber base, revenue streams and profit margins. With the billions carriers are shelling out to make it all happen, there’s real concern over whether the costs will be justified. Will customers give all of their communications dollars to telcos, rather than cable operators or other competitors? Currently, the estimates aren’t promising.
“We’re basically in the 5 to 9 million subscriber range, in terms of how many subscribers telcos will have as of 2010, and that number is underwhelming. If by 2010 they aren’t sporting 10 to 15 million subscribers, I have to believe that’s a disappointment,” says Robin Knight, executive director at Agilent Technologies.
While the telcos are publicly very excited and optimistic about their future in multimedia services, staff members with several key suppliers report they have witnessed a senior executive with at least one of the leading carriers fears that expressing fears that if its new initiative isn’t profitable within two years, the carrier will be going to the capital markets just to make payroll. Such is the danger, and the promise, of IMS.
Author’s note: Alcatel declined to comment for this story. SBC and Telus did not reply to requests for comments and responses.