Despite the promise of high speed services associated with evolving 3G and 4G networks, the challenge of lowering the cost to serve the first subscriber remains a fundamental issue for both carriers and Network Equipment Providers (NEPs). This challenge is particularly acute for carriers building networks in emerging markets or those markets where entrenched competitors can effectively limit the available capital
for new build outs. The target market size for these new revenue opportunities is often less than 50,000 subscribers. Since these new networks are all typically built around an IMS core, the cost to serve the first subscriber of an IMS (IP Multimedia Subsystem) network needs to be examined as one of the first steps in attacking this fundamental cost challenge.
One of the virtues of IMS networks is the rich set of standardized interfaces between all the major IMS components that provide carriers more choice at lower prices than the legacy, vendor specific, proprietary networks IMS is starting to replace. However,
one implication of IMS’ emphasis on standards is the fact that entry level IMS networks require many different network elements before the first subscriber can be served. For instance, even an entry level network still requires three CSCF (Call Session Control Function) functions (proxy, interrogating, serving), an HSS (Home Subscriber Server), various gateways, network/element management systems, and various application servers before service can be provided to the first subscriber. More often than not, each of these network elements requires a pair of carrier-grade servers that hosts the vendor’s IMS software in active/standby configuration. Each IMS network element typically equates to, at least, a pair of under utilized, carrier grade servers. This typically requires the carrier building a new IMS network to acquire
20+ servers for the 10+ IMS network elements needed to provide service to the new carrier’s first subscriber. While initially a major CapEx challenge to the carrier, over time the cost to power, cool, administer and maintain all these disparate servers
quickly becomes an OpEx issue as well.
Traditionally, legacy 2G and 3G networks were designed to scale up: the design challenge to the NEP was to cost effectively scale to millions of subscribers. Today, however, with virtually all major markets covered with several wireless networks, the greenfield opportunities for revenue are typically associated with carriers seeking to enter smaller emerging markets or enter established markets with new services (either 4G or enhanced 3G). In either case, the capital available to carriers to build new networks is significantly less than what was available several years ago at the dawn of 3G. Dominant carriers have, for the first time, begun to partner to build new wireless 3G infrastructure that competing carriers can share, thereby minimizing the capital either carrier needs to raise. In addition, as advanced services are expanded to second and third tier markets, the need to support core network functions in smaller networks puts a severe strain on capital needed to start up these deployments.
Since the greatest opportunity for revenue growth now presents itself in the form of smaller markets with less than 50,000 subscribers, lowering the cost dramatically to serve the first subscriber and to thereby lower the breakeven point in the carrier’s business case becomes the carrier’s first and foremost challenge. To satisfy these new market requirements, NEPs must re-think scalability; low- end scalability – the cost to serve the first subscriber – is now the fundamental engineering challenge facing the industry.
Platform Implications of Low-End Scalability
To effectively address this low-end scalability challenge, platform technology must be developed that offers NEPs and IMS application vendors architectural options that do not require the redesign of all the existing IMS applications. For instance, multi-core processors coupled with powerful virtualization technology that enables the radical consolidation of all the physically discrete carrier-grade servers that today are required to build an IMS core network are a very attractive platform for low-end scalability. If processor cores can replace discrete servers with no change to the existing
IMS application, a consolidated computing platform could effectively host all the IMS core network applications that today require 20+ physically discrete servers. Thus a pair of ATCA blades or a pair of carrier-grade rack mount servers could provide the platform for radically consolidating all the applications in an IMS core network. Replacing 20+ carrier-grade servers with either 2 ATCA blades or 2 carrier-grade servers
